The Influence of Prescribed Fire and Mechanical Fuels Mastication on Soil CO2 Efflux Rates in Two Southeastern U.S. Pine...

MISSING IMAGE

Material Information

Title:
The Influence of Prescribed Fire and Mechanical Fuels Mastication on Soil CO2 Efflux Rates in Two Southeastern U.S. Pine Ecosystems
Physical Description:
1 online resource (237 p.)
Language:
english
Creator:
Godwin, David Robert
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Forest Resources and Conservation
Committee Chair:
Kobziar, Leda Nikola
Committee Members:
Martin, Timothy A
Long, Alan J
Robertson, Kevin
Grunwald, Sabine

Subjects

Subjects / Keywords:
carbon -- field -- fire -- flatwoods -- florida -- flux -- fuel -- mastication -- mechanical -- old -- osceola -- prescribed -- research -- respiration -- soil -- station -- stoddard -- tall -- timbers
Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre:
Forest Resources and Conservation thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Soil CO2 efflux (Rs) is a significant flux of carbon dioxide from ecosystem soils to the atmosphere and is a critical component of the total ecosystem carbon budget. Rs fluxes are comprised of autotrophic (Ra) sources of CO2 produced by plant roots and associated rhizosphere fungi and heterotrophic (Rh) sources of CO2 produced by aerobic soil microbes. A variety of forest management activities, including prescribed fire and mechanical fuels mastication treatments have been shown to significantly influence Rs rates in forests of the Western United States (US), yet these relationships are not well known for southeastern US forests. Prescribed fire is one of the most prevalent forest management tools employed in the Southeast and mechanical fuels treatments are becoming more common in the region as efforts to mitigate potential wildfire behavior in the wildland urban interface grow.  Given that many of these forests provide habitat for endangered species, understanding the implications of management activities on ecosystem carbon dynamics may allow landowners to capitalize on future alternative revenue streams for carbon sequestration services while maintaining their properties in conserved states. This study investigated the influence of prescribed fire and mechanical fuels mastication treatments on Rs rates in longleaf / slash pine flatwoods forests and loblolly / shortleaf pine old-field forests in North Florida, USA.  In the old-field forests, sites managed with over 60-years of annual and biennial dormant season prescribed fire had significantly lower monthly mean Rs rates and estimated annual soil carbon fluxes than sites on which fire was excluded during the same period.  Experimental litter manipulations in the old-field forests found that Rs rates in frequently burned sites increased significantly following litter additions, while sites excluded from did not respond to litter additions.  In the flatwoods forests, neither mechanical fuels treatments nor prescribed fire significantly altered monthly mean Rs rates. These results highlight some of the ways that forest management practices can influence Rs rates.  Our results suggest that future methods to model soil carbon fluxes in the region should incorporate not only current vegetative conditions, but also land management activities and tenure.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by David Robert Godwin.
Thesis:
Thesis (Ph.D.)--University of Florida, 2012.
Local:
Adviser: Kobziar, Leda Nikola.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-12-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0044618:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 THE INFLUENCE OF PRESCRIBED FIRE AND MECHANICAL FUELS MASTICATION ON SOIL CO2 EFFLUX RATES IN TWO SOUTHEASTERN U.S. PINE ECOSYSTEMS By DAVID ROBERT GODWIN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY O F FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 David Robert Godwin

PAGE 3

3 To my parents, grandparents, and family for the encouragement, inspiration, and support they have selflessly provided my entire life. To my patient and wonderful wife for her su pport through this long process and to our young son who brings such joy to each and every day

PAGE 4

4 ACKNOWLEDGMENTS This study would not have been possible without the encouragement, support, and guidance of my major professor, Dr. Leda Kobziar and supervisory committee members: Dr. Sabine Grunwald, Dr. Alan Long, Dr. Tim Martin, and Dr. Kevin Robertson. This research was funded in part by a Graduate Resear ch Innovation (GRIN) grant from the Joint Fire Science Program and the Association for Fire Ecology The Influence of Prescribed Fire and Understory Fuels Mastication on Soil Carbon Respiration Rates Significant support coordinating and establ ishing the Osceola study sites came from University of Florida Fire Science Lab alumn us Dr. Jesse Kreye. Additional support in hiring and managing field technicians was provided by University of Florida Fire Science Lab alumn us Dr. Adam Watts. Much of th e f ield data at the Osceola study site and elsewhere were collected through the tireless assistance of University of Flori da Fire Science Lab technicians, students and volunteers : Michael Camp, Dawn McKinstry, Marissa Streifel, and Alex Kattan This stud y would not have been possible without the cooperation of the research site coordinators: Dr. Kevin Robertson and Dr. Ron Masters of the Tall Timbers Research Station, Dr. Alan Long, Dr. Michael Andreu, Dan Schultz, and Gary Johns of the University of Flor ida Austin Cary Forest, and Peter Myers, Fire Management Officer of the Osceola National Forest USDA Forest Service Finally, I thank my loving wife, brother, parents and grandparents for their patience, encouragement and dedication towards the c ompletion of this study.

PAGE 5

5 TABLE OF CONTENTS p age ACKNOWLEDGMENTS ................................ ................................ ................................ 4 LIST OF TABLES ................................ ................................ ................................ ........... 7 LIST OF FIGURES ................................ ................................ ................................ ...... 10 LIST OF ABBREVIATIONS ................................ ................................ .......................... 14 ABSTRACT ................................ ................................ ................................ .................. 15 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ... 17 2 FORTY YEARS OF PRESCRIBED FIRE ALTERS SOI L CO2 EFFLUX RATES AT THE STODDARD FIRE PLOTS IN NORTH FLORIDA ................................ ..... 22 Background ................................ ................................ ................................ ........... 22 Methods ................................ ................................ ................................ ................ 26 Study Site ................................ ................................ ................................ ........ 26 Sampling ................................ ................................ ................................ ......... 28 Analysis ................................ ................................ ................................ ........... 30 Results ................................ ................................ ................................ .................. 32 Discussion ................................ ................................ ................................ ............. 38 Conclusion ................................ ................................ ................................ ............. 50 3 THE EFFECTS OF LITTER INPUTS AND PRESCRIBED FIRE ON SOIL CO2 EFFLUX RATES IN NORTH FLORIDA OLD FIELD FORESTS ............................ 80 Background ................................ ................................ ................................ ........... 80 Methods ................................ ................................ ................................ ................ 83 Study Site ................................ ................................ ................................ ........ 83 Litter Manipula tion and Sampling ................................ ................................ .... 84 Analysis ................................ ................................ ................................ ........... 88 Results ................................ ................................ ................................ .................. 89 Analysis of Treatment Effects ................................ ................................ .......... 91 Effects of Treatments o n the Response of R s to Abiotic Factors ...................... 92 Discussion ................................ ................................ ................................ ............. 94 Effect of Prescribed Fire Management ................................ ............................ 95 Effect of Litter Addition ................................ ................................ .................... 96 Effect of Litter Exclusion ................................ ................................ .................. 99 Importance of Soil Temperature ................................ ................................ .... 101 Importance of Soil Moisture ................................ ................................ ........... 104 Conclusions ................................ ................................ ................................ ......... 105

PAGE 6

6 4 THE INFLUENCE OF PRESCRIBED FIRE AND UNDERSTORY FUELS MASTICATION ON SOIL CO2 EFFLUX RATES IN TWO NORTH FLORIDA FLATWOODS FORESTS ................................ ................................ .................... 140 Background ................................ ................................ ................................ ......... 140 Methods ................................ ................................ ................................ .............. 143 Study Areas ................................ ................................ ................................ .. 143 Sampling ................................ ................................ ................................ ....... 145 Field Measurements ................................ ................................ ...................... 146 Analysis ................................ ................................ ................................ ......... 149 Results ................................ ................................ ................................ ................ 152 Treatment Effects ................................ ................................ .......................... 152 Overall Drivers of Soil CO 2 Efflux ................................ ................................ .. 156 Treatment Specific Drivers of Soil CO 2 Efflux ................................ ................ 157 Seasonal Drivers of Soil CO 2 Efflux ................................ ............................... 159 Multiple Regression Models ................................ ................................ .......... 160 Temperature Response ................................ ................................ ................. 162 Estimated Carbon Flux ................................ ................................ .................. 163 Discussion ................................ ................................ ................................ ........... 164 Effects of Prescribed Fire and Mechanical Fuels Mastication ........................ 165 Soil CO 2 Efflux Response to Temperature Fluctuations ................................ 170 S oil CO 2 Efflux Response to Soil Moisture and Precipitation ......................... 175 Effects of Treatment on Soil Carbon Flux ................................ ...................... 176 Conclusion ................................ ................................ ................................ ........... 177 5 SUMMARY AND SYNTHESIS ................................ ................................ ............. 216 Summary ................................ ................................ ................................ ............. 216 Synthesis ................................ ................................ ................................ ............. 219 LIST OF R EFERENCES ................................ ................................ ............................ 225 BIOGRAPHICAL SKETCH ................................ ................................ ......................... 237

PAGE 7

7 LIST OF TABLES Table page 2 1 Param CO 2 efflux rates at the Tall Timbers Research Station, FL ................................ 52 2 2 Mean forest characteristics per prescribed fire treatm ent type at the Tall Timbers Research Station, FL ................................ ................................ ........... 53 2 3 R epeated measures ANOVA for soil CO 2 efflux, soi l temperature and soil moisture content at the Tall Timbers Research Station, FL ............................... 54 2 4 Mean seasonal and total study period soil temperature per prescribed fire treatment type at the Tall Timbers Research Station, FL ................................ ... 55 2 5 Mean seasonal and total study period soil moisture content per prescribed fire treatment type at the Tall Timbers Research Station, FL ............................. 56 2 6 Mean seasonal and total study period soil C O 2 efflux per prescribed fire treatment type at the Tall Timbers Research Station, FL ................................ ... 56 2 7 2 efflux, soil temperature soil moisture content and plot ve getative and meteorological characteristics .......... 57 2 8 Linear regression relationships between soil CO 2 efflux rates and field conditions by fire return interval ................................ ................................ ......... 58 2 9 Results of nonlinear models of soil CO 2 efflux rates using soil temperature as a predictor ................................ ................................ ................................ ......... 59 2 10 Linear regression relationship between soil CO 2 efflux rate s and soil temperature by fire return interval and season ................................ .................. 60 2 11 Seasonal nonlinear models of soil CO 2 efflux rates using soil temperature as a predictor ................................ ................................ ................................ ......... 61 2 12 Seasonal linear models of soil CO 2 efflux rates using soil moisture content as a predictor ................................ ................................ ................................ ......... 62 2 13 Step wise multiple regression models to explain soil CO 2 effl ux rates using field parameters ................................ ................................ ................................ 63 3 1 Plot level variables investigated for their influence on soil CO 2 efflux rates at the Tall Timbers Research Station, FL ................................ ............................ 107 3 2 Mean forest characteristics per prescribed fire treatment type at the Tall Timbers Research Station, FL ................................ ................................ ......... 108

PAGE 8

8 3 3 R epeated measures ANOVA for soil CO 2 efflux, soil temperature, and soil moisture content means at the Tall Timbers Research Station, FL .................. 1 09 3 4 Mean soil CO 2 efflux, soil temperature, and soil moisture content by prescribed fire treatmen t at the Tall Timbers Research Station ....................... 110 3 5 Mean s oil CO 2 efflux, soil temperature, and soil moisture content for the entire study period per litter treatment at the Tall Timbers Research St ation ... 110 3 6 R epeated measures ANOVA for soil CO 2 efflux, soil temperature, and soil moisture content within treatments at th e Tall Timbers Research Station ........ 111 3 7 Soil CO 2 efflux, soil temperature, and soil moisture content by litter treatment and fire return interval for the Tall Timbers Research Station .......................... 112 3 8 Linea r regression models of the relationships between soil CO 2 efflux rates and soil temperature by fire return interval and litter treatment type ................ 113 3 9 No n linear exponential models of soil CO 2 e fflux rates (R s ) and soil temperature by fire return interval and litter treatment type r ........................... 114 3 10 Linear regression of soil temperature and monthly mean ambient air temperature by litter treat ment type and fire return interval .............................. 115 3 11 Linear regression of the relationships between soil CO 2 efflux rates and soil moisture content by litter treatment type and fire return interval ...................... 116 3 12 Linear regression of the relationships between soil CO 2 efflux rates and monthly precipitation by litter treatment type and fire return interval ................ 117 3 13 Linear regression of soil moisture content and monthly precipitation by litter treatment type and fire return interval ................................ .............................. 118 3 14 Linear regression of soil temperatur e and monthly precipitation by litter treatment type and fire return interval ................................ .............................. 119 4 1 Parameters accessed for influence on soil CO 2 efflux rates at the Austin Cary Forest and Osceola National F orest, Florida, USA ................................ .......... 179 4 2 Mean forest characteristics per treatment at the Austin Cary Forest and Osceola National Forest, Florida, USA. ................................ ........................... 181 4 3 R epeated measures ANOVA for soil CO 2 efflux, soil temperature, and soil moisture content at the Austin Cary Forest and Osceola National Forest ........ 182 4 4 Overall means of soil tem perature, moisture content, and soil CO 2 efflux rates per treatment and study site ................................ ................................ ... 183

PAGE 9

9 4 5 Dormant and growing season soil temperature, soil moisture, and soil CO 2 efflux at the Austin Cary For est and Osceola National Forest .......................... 183 4 6 2 efflux, soil temperature, soil moisture contentand field conditions for the Osceola National Forest ....... 184 4 7 among soil CO 2 efflux rates and field conditions for the Austin Cary Forest ................................ ............................... 185 4 8 S imp le linear regression models of soil CO 2 efflux rates and field conditions by study area and treatment ................................ ................................ ............ 186 4 9 N onlinear models of soil CO 2 efflux rates using soil temperature as a predictor ................................ ................................ ................................ .......... 187 4 10 S imple linear regression models of soil CO 2 efflux rates and field conditions by study area, treatment, and season ................................ ............................. 188 4 11 S eason specific nonlinear models of soil CO 2 efflux rates and soil temperature at the Austin Cary Forest and Osceola National Forest ............... 190 4 12 Step wise multiple linear regression models by stud y site and treatment predicting soil CO 2 efflux rates from field parameters ................................ ...... 191 4 13 Step wise multiple linear regression models by study site, treatment, and season predicting soil CO 2 efflux rates from field parameters .......................... 192

PAGE 10

10 LIST OF FIGURES Figure page 2 1 The research site, Tall Timbers Research Station, was located in Leon County, Florida, USA ................................ ................................ ......................... 64 2 2 Ground and aerial images of three of the soil CO 2 efflux sampling plots located within the Tall Timbers Research Station ................................ .............. 65 2 3 PVC soil CO 2 efflux sampling collar in stalled at a frequently burned plot at the Tall Timbers Research Station ................................ ................................ .... 66 2 4 Monthly soil temperature and monthly 2 m air temperature for prescribed fire treatment types at the Tall Timbers Research Station ................................ ....... 67 2 5 Mean soil CO 2 efflux, soil temperature, and soil moisture content per treatment type at Tall Timbers Research Station ................................ ............... 68 2 6 Monthly soil moisture content (m 3 /m 3 ) for three prescribed fire treatment types at the Tall Timbers Research Station ................................ ....................... 69 2 7 Monthly regional Palmer Drought Severity Index (PD SI) scores from the National Oceanic and Atmospheric Administration (NOAA) ............................... 70 2 8 Monthly soil CO 2 efflux rates for three prescribed fire treatment types at the Tall Timbers Research Station ................................ ................................ .......... 71 2 9 Linear regression of soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and site biotic factors at the Tall Timbers Research Station ................................ ............ 72 2 10 Linear regression of soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and litter and duff depth at the Tall Timbers Research Station ................................ ......... 73 2 11 Linear regression of soil CO 2 efflux rat es and soil temperaturefor three prescribed fire intervals at the Tall Timbers Research Station ........................... 74 2 12 The relationship between soil CO 2 efflux rates and soil temperature as modeled using an e xponential equation. ................................ ........................... 75 2 13 Linear regression of soil CO 2 efflux and monthly air temperature for prescribed fire intervals at the Tall Timbers Research Station ........................... 76 2 14 R elationship between soil CO 2 efflux rates and air temperature modeled using an exponential equation ................................ ................................ ........... 77 2 15 Linear regression of soil CO 2 efflux and mont hly mean soil moisture content for prescribed fire intervals at the Tall Timbers Research Station ...................... 78

PAGE 11

11 2 16 Predicted monthly total soil carbon flux from August 2009 to July 2010 by prescribed fi re treatment ................................ ................................ .................... 79 3 1 Map of the study area at the Tall Timbers Research Station in Leon County, Florida, USA. ................................ ................................ ................................ ... 120 3 2 Ground and ae rial images of three of the soil CO 2 efflux sampling plots located within the Tall Timbers Research Station ................................ ............ 121 3 3 Photograph of a 20 cm soil CO 2 efflux sample collar and 0.16 m 2 treatment box and litter exclusion at the Tall Timbers Research Station .......................... 122 3 4 Photograph of the LICOR Biosciences LI 8100 soil CO 2 efflux sampling instrument with soil moisture and temperature probes. ................................ .... 123 3 5 Plot of seven months of air temperature records and precipitation for the year 2011 for a siteapproximately 30 km from Tall Timbers Research Station ........ 124 3 6 Plot of seven months of monthly Palmer Drought Severity Index values for the year 2011 from the National Oceanic and Atmospheric Administration. ..... 125 3 7 M ean soi l CO 2 efflux rates, soil temperature, and soil moisture content by litter treatment ................................ ................................ ................................ 126 3 8 Monthly mean soil CO 2 efflux rates, soil temperature, and soil moisture content by prescribed fire t reatment ................................ ................................ 127 3 9 Monthly mean soil CO 2 efflux rates by litter treatment and fire treatment type 128 3 10 Overall mean soil CO 2 eff lux rates by litter treatment within fire each treatment type ................................ ................................ ................................ 129 3 11 M ean soil moisture content by litter and fire treatment typ e ............................. 130 3 12 Overall mean soil moisture content by litter manipulation treatment within each fire treatment type ................................ ................................ .................. 131 3 13 Monthly mean soil temperature by litter manipulation and fire treatment type .. 132 3 14 Overall mean soil temperature by litter manipulation treatment within each fire treatment type ................................ ................................ .......................... 133 3 15 Linear regr ession of soil CO 2 efflux and soil temperature for three litter treatment types within the 1YR prescribed fire interval ................................ .... 134 3 16 Linear regression of soil CO 2 efflux rates and soil temperature for three litter treatment types within the 2YR prescribed fire interval ................................ .... 135

PAGE 12

12 3 17 Linear regression of monthly mean soil CO 2 efflux rates and soil temperature for three litter treatment types w ithin the 40YR prescribed fire interval ............ 136 3 18 Exponential model of the r elationship between soil CO 2 efflux an d soil temperature in the 1YR prescribed fire treatment ................................ ............ 137 3 19 Exponential model of the relationship between soil CO 2 efflux and soil temperature in the 2YR prescribed fire treatment ................................ ............ 138 3 20 Exponential mode l of the relationship between soil CO 2 efflux and soil temperature in the 40YR prescribed fire treatment ................................ .......... 139 4 1 Map of the study areas at the Osceola National Forest near Lake City, Florida an d Austin Cary Forest near Gainesville, Florida, USA ........................ 193 4 2 Examples of the four pine flatwoods forest management types sampled in the Osceola National Forest study site near Lake City, Florida USA .................... 194 4 3 Pine flatwoods forest management types represented in the study at the Austin Cary Memorial Forest, Gainesville, Florida, USA.. ................................ 195 4 4 Monthly mean soil CO 2 efflux rates, soil temperature and soil moisture content per treatment at the Osceola National Forest ................................ ...... 196 4 5 Monthly mean soil CO 2 efflux rates soil temperature and soil moisture content per treatment at the Austin Cary Forest ................................ .............. 197 4 6 Monthly Palmer Drought Severity Index values for the region containing the Austin Cary Forest and the Osc eola National Forest study areas .................... 198 4 7 Treatment means of soil CO 2 efflux rates soil temperature and soil moisture contentfor the Osceola National Forest. ................................ .......................... 199 4 8 Treatment means of soil CO 2 efflux soil temperature and soil moisture for the Austin Cary Forest ................................ ................................ ..................... 199 4 9 Linear regressions of the relationships between s oil CO 2 efflux and biotic and abiotic factors for the Osceola National Forest ................................ ................ 200 4 10 Linear regression of the relationships between soil CO 2 efflux rates and biotic and abiotic factors at the Osceola National Forest ................................ .......... 201 4 11 Linear regression of the relationships between soil CO 2 efflux rates and multiple biotic and abiotic factors at the Austin Cary Forest ............................. 202 4 12 Linear regression of the relationships between mean soil CO 2 efflux rates and duff and litter depth for the Austin Cary Forest. ................................ ......... 203

PAGE 13

13 4 13 Linear r egression of the relationships between monthly mean soil CO 2 efflux rates and soil temperature at the Osceola National Forest .............................. 204 4 14 Linear regression of the relationships between monthly mea n soil CO 2 efflux rates and soil temperature at the Austin Cary Forest ................................ ....... 205 4 15 Non linear regression of the relationships between monthly mean soil CO 2 efflux rates and soil temperature at the Osceola National Forest ..................... 206 4 16 Non linear regression of the relationships between monthly mean soil CO 2 efflux rates and soil temperature at the Austin Cary Forest. ............................ 207 4 17 Seasonal (dormant and growing) linear regressions of soil CO 2 efflux rates and soil temperature at the Osceola National Forest ................................ ....... 208 4 18 Seasonal ( dormant and growing ) non linear regressions of soil CO 2 efflux rates and soil temperature at the Osceola National Forest .............................. 209 4 19 Seasonal (dormant and growing) linear regressions of soil CO 2 e fflux rates and soil temperature. ................................ ................................ ....................... 210 4 20 Seasonal (dormant and growing ) non linear regressions of monthly mean soil CO 2 efflux rates and soil temperature ................................ ............................. 211 4 21 Predicted monthly total soil carbon flux for the four treatments at the Osceola National Forest. ................................ ................................ ............................... 212 4 22 Predicte d annual total soil carbon flux for the fou r treatments at the Osceola National Forest. ................................ ................................ ............................... 213 4 23 Predicted monthly total soil carbon flux for the two prescribed fire treatments at the Austin Cary Forest. ................................ ................................ ................ 214 4 24 Predicte d annual total soil carbon flux at the Austin Cary Forest. .................... 215 5 1 Conceptual model of the sources and drivers of soil respiration rates in fores ted ecosystems managed with prescribed fire. ................................ ....... 223 5 2 Conceptual model of the influence of prescribed fire frequency and seasonal and environmental fact ors on soil respiration rat e ...................... 224

PAGE 14

14 LIST OF ABBREVIATION S FRI Fire return interval (yr 1 ) M s Soil volumetric moisture content (m 3 x m 3 ) R a Autotrophic soil respiration ( 2 m 2 sec 1 ) R h Heterotrophic soil respiration ( 2 m 2 sec 1 ) R s Soil respiration ( 2 m 2 sec 1 ) T s Soil temperature (C)

PAGE 15

15 Abstract of Disserta tion Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE INFLUENCE OF PRESCRIBED FIRE AND MECHANICAL FUELS MASTICATION ON SOIL CO2 EFFLUX RATES IN TWO SOUTHEASTERN U.S. PINE ECOSYSTEMS By David Robert Godwin December 2012 Chair: Leda Kobziar Major: Forest Resources and Conservation Soil CO 2 efflux (R s ) is a significant flux of carbon dioxide from ecosystem soils to the atmosphere and is a critical component of the total ecosystem carbon budget R s fluxes are comprised of autotrophic (R a ) sources of CO 2 produced by plant roots and associated rhizosphere fungi and heterotrophic (R h ) sources of CO 2 produced by aerobic so il microbes A variety of forest management activities, including prescribed fire and mechanical fuels mastication treatments have been shown to significantly influence R s rates in forests of the Western United States (US) yet these relationships are not well known for s outheastern US forests. Prescribed fire is one of the most prevalent forest management tools employed in the s outheastern US, and mechanical fuels treatments are becoming more common in the region as efforts to m itigate potential wildfire behavior in the wildland urban interface grow. Given that many of these forests provide habitat for endangered species, understanding the implications of management activities on ecosystem carbon dynamics may allow landowners to capitalize on future alternative revenue streams for carbon sequestration services while maintain ing their properties in conserved states

PAGE 16

16 This study investigated the influence of prescribed fire and mechanical fuels mastication treatments on R s rates in longleaf / slash pine flatwoods forests and loblolly / shortleaf pine old field forests in N orth Florida, USA. In the old field forests sites managed with over 60 years of annual and biennial dormant season prescribed fire had significantly l ower monthly mean R s rates and estimated annual soil carbon flux es than sites on which fire was excluded during that same period. Experimental litter manipulations in the old field forests found that R s rates in frequently burned sites increased significa ntly following litter additions while sites excluded from fire for over 60 years did not respond to litter additions In the flatwoods forests neither mechanical fuels treatments nor prescribed fire significantly alter ed monthly mean R s rates These res ults highlight some of the ways that forest management practices can influence R s rates Our results suggest that future methods to model soil carbon emissions in the region should incorporate not only current vegetative conditions, but also land manageme nt activities and tenure.

PAGE 17

17 CHAPTER 1 INTRODUCTION Managing forests to increase carbon sequestration and decrease carbon emissions has been suggested as a method for reducing global atmospheric CO 2 concentrations that have increased over the last century ( IPCC, 1995; Lal, 2005; Woodbury et al., 2007) As the southeastern United States (US) has over 81 million ha of forested land there exists significant potential for public and private forest landowner compensation for carbon sequestration ser (USFS/FIA 2006 ; Maier et al., 2012 ) Furthermore, given that many of the forested lands in the southeastern US provide valuable habitat for various threatened and endangered and sport hunting species alternative revenue s treams for carbon sequestration services may facilitate some landowners maintain ing their properties in conserved states (Engstrom and Palmer, 2005) It has been suggested that understanding soil carbon pools and fluxes are the weakest link s in assessing c arbon in s outheastern US forests (Raich and Schlesinger, 1992; Johnson et al., 2001). This is important as m uch of the carbon sequestered in temperate forested systems is found within the soils (50 60%) with soil CO 2 efflux comprising a significant porti on (50 60%) of temperate forest total ecosystem carbon budgets (Post et al., 1982; Raich and Schlesinger, 1992 ; Lal, 2005; Noormets et al. 2010 ) A t landscape scales, g i ven that soils contribute such large flux es to atmospher ic CO 2 even small changes in soil CO 2 efflux rates over broad regions could have significant impacts on overall atmospheric CO 2 concentrations (Raich and Schlesinger, 1992 ; Bond Lamberty and Thomson, 2010 ) Because of the importance of soil CO 2

PAGE 18

18 ef flux rates in local, regional, and global carbon cycles it is important to understand how forest management practices influence soil CO 2 efflux rates. Soil CO 2 efflux (R s ) is a combination of CO 2 respired by plant roots and associated rhizosphere fungi (R a ) and heterotrophic soil microorganisms (R h ) (Luo and Zhou, 2006; Subke et al., 2010) Soil CO 2 efflux is the product of a multitude of interrelated biogeochemical factors that govern the production of CO 2 by plant roots (and associated mycorrhizal fungi) and soil micro and macro biota, including: soil temperature, soil moisture content, aboveground vegetative composition and belo w ground carbon allocation, phenology, soil carbon and nutrient content and disturbance processes (Raich and Tufekc iogul, 2000; Ryan and Law, 2005 ; Luo and Zh ou, 2006 ) In addition a suite of physical factors including soil porosity, CO 2 pressure gradients, surface wind speed, and surface air turbulence influence the evolution of CO 2 to the soil surface (Luo and Zhou, 2006) Presc ribed fire is one of the most prevalent forest management tools employed in the management of conserved lands in the southeastern US with over 2.4 million ha burned in 2011 (Waldrop and Goodrick, 2012) Prescribed fire is frequently used to maintain open, low density pine forests that favor game and non game wildli fe species and fire dependent plant species as well as to reduce wildfire risk by consuming litter, understory, and midstory vegetative fuels (Outcalt and Wade, 2004; Mitchell et a l., 2006 ; Waldrop and Goodrick, 2012 ) While prescribed fire in the southeastern US is an obvious source of atmospheric carbon in the short term, investigations of the rapid response of vegetation following prescribed fires suggest that the ecosystem reco very and sequestration of carbon lost via emissions is relatively fast (1 2 years) especially in

PAGE 19

19 comparison to other regions of the US (Wiedinmyer and Neff, 2007; Lavoie et al., 2010). It is known that variations in the frequency or season of pr escribed fire management regimes can result in significant changes in forest vegetation structure and composition in the southeastern US (White et al., 1990; Waldrop et al., 1992; Knapp et al., 2009; Glitzenstein et al., 2012). What are not s o well known are the ef fects of such management regime variations on the biotic and abiotic factors that drive forest soil carbon fluxes. Mechanical fuels mastication treatments are becoming more common in the southeastern US as wildfire prone forests and urban areas intermix (Mitchell et al., 2006; Menges and Gordon, 2010) Mechanical fuels mastication is used to reduce understory fuel heights which has been shown to reduce wild fire behavior in many systems (Agee and Skinner, 2005; Glitzenstein et al., 20 06; Kobziar et al., 2009 ; Kreye, 2012 ) In many wildland urban interface areas in the southeastern US, prescribed fire has become difficult for land managers to implement due to concerns from adjacent and nearby landowners over smoke and wild fire risk and as such, many land managers and agencies are opting to use mechanical fuels treatments in place of prescribed fire (Miller and Wade, 2003; Long et al., 2004 ; Menges and Gordon, 2010 ) As the implementation o f mechanical fuel treatments has i ncreased, it is important to understand the influence of such practices on forest carbon dynamics Previous studies of mechanical fuels mastication in Western US systems have shown that treatments can significantly alter soil CO 2 efflux rates (Kobziar and Stephens, 2006 ; Ryu et al., 2009 ) as well as soil environmental factors known to influence long term soil carbon dynamics (Concilio et al., 2005; Kobziar and Stephens, 2006; Xu et al., 2011) Given the increased use of

PAGE 20

20 mechanical fuels mastication treatm ents in place of or in combination with prescribed fire in the southeastern US it is important to understand the effects of such activities on forest soil carbon fluxes and soil abiotic and biotic conditions. Soil CO 2 efflux has been signific antly positively correlated with soil temperature in many ecosystems and it has been suggested that increases in soil temperature due to global climate change may drive landscape level elevated soil CO 2 efflux rates and soil carbon loss es ( Raich and Schles inger, 1992; Ryan and Law, 2005 ; Bond Lamberty and Thomson, 2010 ). Experimental manipulations simulating future climate change scenarios have also found that soil CO 2 efflux rates can increase following exposure to elevated atmospheric CO 2 concentrations (Schlesinger and Andrews, 2000; Butnor et al., 2003; Carney et al., 2007) While much remains to be understood regarding the interactions of climate change and ecosystem carbon cycles (Bonan, 2008), t he results of these studies further reinforce the impor tance of understanding the implications of forest management practices on soil CO 2 efflux rates A s mentioned previously a variety of forest management activities, including prescribed fire and mechanical fuels mastication treatments have been shown to si gnificantly influence soil CO 2 efflux rates in the W estern US yet these relationships are not well known for southeastern US forests (Concilio et al., 2005; Tang et al., 2005; Kobziar and Stephens, 2006; Ryu et al., 2009; Xu et al., 2011) T o address this knowledge gap, t he following studies in this document sought to improve the understand ing of forest management practices on soil CO 2 efflux rates in two forest types managed for conservation in Florida, USA. The objectives of these studies were to 1 ) assess the implications of prescribed fire management regimes on monthly mean

PAGE 21

21 and annual soil CO 2 efflux rates in old field forests ; 2 ) assess the importance of aboveground litter inputs in influencing soil CO 2 efflux rates for a range of old fi eld forest prescribed fire management regimes; and 3 ) evaluate the effects of mechanical fuels mastication treatments, prescribed fire, and mechanical fuels mastication treatments followed by prescribed fire on soil CO 2 efflux rates and soil abiotic condit ions in pine flatwoods forests.

PAGE 22

22 CHAPTER 2 FORTY YEARS OF PRESC RIBED FIRE ALTERS SO IL CO2 EFFLUX RATES AT THE STODDARD FIRE PLOTS IN NORTH FLORIDA Background Soil CO 2 efflux (R s ) is a significant flux of carbon dioxide from ecosystem soils to the greater atmosphere and is a critical component in determining total ecosystem carbon budgets (Raich and Schlesinger, 1992; Schlesinger and Andrews, 2000; Ryan and Law, 2005) It has been estimated that soil CO 2 efflux represents one of the largest gl obal terrestrial fluxes of carbon to the atmosphere, with the total annual R s flux (75 Pg C yr 1 ) an order of magnitude greater than current annual anthropogenic C emissions from fossil fuel combustion (6 Pg C yr 1 ) (Luo and Zhou, 2006) Soil CO 2 efflux i s comprised of autotrophic (R a ) and heterotrophic (R h ) sources of CO 2 (Luo and Zhou, 2006; Subke et al., 2010) Total R s is a function of a multitude of interrelated biogeochemical factors that govern the production of autotrophic soil CO 2 eff lux by plant roots (and associated mycorrhizal fungi) and heterotrophic soil CO 2 efflux by soil micro and macro biota, including: soil temperature, soil moisture content, aboveground vegetative composition and carbon allocation, phenology, soil carbon cont ent and disturbance processes (Raich and Tufekciogul, 2000; Ryan and Law, 2005) Beyond the biological factors associated with soil CO 2 efflux, a suite of physical factors including soil porosity, CO 2 pressure gradients, surface wind speed, and surface a ir turbulence govern the R s rate at the soil surface (Luo and Zhou, 2006) Given the magnitude of this source of CO 2 research over the past decades has increased to improve the understanding of the drivers and mechanisms influencing R s (Luo and Zhou, 2006 ). While many of the significant factors influencing R s rates have been identified, much remains to be understood regarding the effect of specific land

PAGE 23

23 management practices on R s rates and those driving factors (Schlesinger and Andrews, 2000; Ryan and Law 2005; Luo and Zhou, 2006) Studies have shown that both fire and forest management can influence soil carbon pools, biogeochemical properties, and R s rates (Johnson, 2001; Johnson et al., 2002; Certini, 2005; Kobziar and Stephens, 2006, Kobziar, 2007) For example, in a study of an oak forest in Oklahoma, USA, Williams et al. (2012) found that a 20 year management regime of frequent prescribed fire significantly reduced soil organic matter, increased bulk density, and reduced the biomass of certain soil bacteria; all factors that may have led to reduced sources of R h and subsequently overall R s rates Similarly, in a study of a mixed conifer forest in California, USA, Ryu et al. (2009) found that prescribed fire reduced R s rates while simultaneo usly altering soil conditions that would otherwise be associated with increased R s rates. In a contrasting study of forest management and prescribed fire in a mixed conifer forest in California, USA, and an upload oak forest in Missouri, USA, Concilio et al. (2005) found that R s increased in both sites following forest thinning operations and that prescribed burning significantly altered forest floor conditions, but had no clear effect on R s Also in California, Kobziar and Stephens (2006) found that pres cribed fire within a ponderosa pine plantation reduced R s rates in stands treated with understory mechanical mastication, while increasing R s rates in non masticated stands. The results of these studies describe the complex interactions that can occur bet ween prescribed fires, forests, and soil CO 2 efflux rates. Given the frequency of the use of prescribed fire in the upland forests of the southeastern USA, it is notable that few studies have addressed the influence of the management practice on R s rates in the region.

PAGE 24

24 The effect of fire on R s depends on the ecosystem, fire history, and fire severity. F ire can influence R s rates by impacting the R h and R a sources of CO 2 ( Neary et al., 1999; Luo and Zhou, 2006 ). Short term a utotrophic production of CO 2 ca n be reduced by fire due to aboveground plant mortality and injury The long term impact s of fire on R a are variable however R a production often increases with time since fire as vegetation recovers following disturbance In many cases the consumption of vegetation and surface fuels by fire reallocates nutrient resources via incomplete combustion and subsequent deposition of ash, char, and other residues ( Neary et al. 1999; Medvedeff, 2012) In the period following the deposition of those residues bot h plants and soil microbes may respond positively to the availability of su ch resources subsequently increasing R s rates. Changes in forest vegetative composition and structure such as those described by Glitzenstein et al. (2012) caused by long term pre scribed fire management regimes may also impact R h sources of CO 2 by influencing litter and duff quality, production and accumulation rates as s uch factors have been shown to influence soil microbial populations and metabolic activity (Sulzman et al., 2 012). Fire can also reduce R h sources through soil microbial population mortality caused by the direct combustion or heating of litter and duff layers and upper soil horizons (Luo and Zhou, 2006) Prescribed fire is one of the dominant tools for forest m anagement in the old field famous for its history of fire ecology research (Stoddard, 1969; Engstrom and Palmer, 2005; Way, 2006) The loblolly pine shortleaf pin e old field forest type is frequently found on a collection of large privately owned plantation properties managed to promote

PAGE 25

25 Northern Bobwhite quail ( Colinus virginianus (L.)) populations for recreational hunting and selective timber harvesting (Figure 2 1) (Moser, 2002; Engstrom and Palmer, 2005) Many of these properties contain a mixture of old field forests and patches of remnant upland longleaf pine forests that together provide habitat for many threatened and endangered species such as the Red cocka ded Woodpecker ( Picoides borealis ). Collectively, these private properties represent a significant regional holding of conserved lands, totaling over 121,000 ha found north of the Cody Scarp and stretching from the Ochlocknee River to the Aucilla River in Florida and Georgia (Paisley, 1989; Engstrom and Palmer, 2005) Regional suburban development, economic incentives for alternative uses, and changes in property ownership over the past several decades have resulted in the decline in total area managed fo r conservation purposes (Engstrom and Palmer, 2005) Beyond the Red Hills region, across much of the southeastern United States, many former agricultural lands are now covered in similar old field forests comprised of loblolly pine, shortleaf pine, and mi xed hardwoods representing an area estimated to cover up to 21 million ha (Frost, 1993) Future and anticipated alternative revenue streams for carbon sequestration services may provide incentives for old and new landowners in the region to maintain their properties in conserved states providing critical habitat for threatened and endangered species. Previous and ongoing research in old field plots at the Tall Timbers Research Station in Leon County, Florida, USA has quantified abovegroun d carbon pools across multiple prescribed fire management regimes, with the intent of understanding the implications of specific prescribed fire return intervals on forest carbon allocation and dynamics (K Robertson, 2009 pers. comm.).

PAGE 26

26 Given the e ffect s of fire on the autotrophic and heterotrophic sources of soil respiration documented in studies of other ecosystems this study sought to address the following hypothes is for loblolly pine old field forests : a prolonged management regime of frequent prescr ibed fire results in reduced soil CO 2 efflux rates relative to a management regime of fire exclusion In addition, this study sought to understand the influence of biotic and abiotic factors including soil temperature, soil moisture, monthly precipitation and stand characteristics on soil CO 2 efflux rates. Given that previous research in the literature has found th o se factors to influence soil respiration rates, it was hypothesized that variations in those factors would explain any observed differences i n soil respiration rates among the prescribed fire treatments. The intent of this research is to support broader efforts in the region to quantify the effects of prescribed fire on S outheastern forest carbon dynamics and carbon sequestration. Studies su ch as this also provide insight into the response of ecosystem carbon dynamics to forecasted changes in temperature and moisture regimes due to global climate change. Methods Study S ite The study sites were located within the Tall Timbers Fire Ecology Research Plots (Stoddard Fire Research Plots) at the Tall Timbers Research Station (TTRS) in Leon County, Florida, USA, approximately 30 km from the cities of Tallahassee, Florida (to the south) and Thomasv 1) (Clewell and Komarek, 1975; Glitzenstein et al., 2012) The Stoddard Plots established in 1960 as a long term study of fire frequency on old field forest vegetation and soils, have been consi stently managed with a sequence of differing prescribed fire

PAGE 27

27 return intervals for over fifty years (Clewell and Komarek, 1975; Glitzenstein et al., 2012) For this study, sampling took place within the annually burned (1YR), biennially burned (2YR) and fi re excluded (40YR) Stoddard Plots (Figure 2 2). Prior to establishment of the plots in 1960 and 1966, the areas had been burned annually since agricultural abandonment in the late 1800s and 1920s (K Robertson, 2012 pers. comm.). The study sites were loca ted approximately 59 m a.s.l. Average annual precipitation was 137 cm with the majority falling during the summer months of June, July and August (National Climate Data Center 2009, Thomasville, Georgia). Mean maximum and minimum temperatures for January and July for the area from long term records (1971 2000) were 16.8C and 4.6C for January and 33C and 21.8C for July (National Climate Data Center 2009, Thomasville, Georgia). Soils within the sites were heavily cultivated for corn and cotton from the 1820s 1920s and occasionally as recent as the 1950s, with subsequent understory and overstory vegetation assemblages highly influenced by past agricultural practices (Clewell and Komarek, 1975) Soils were generally classified as fine loamy, kaolinitic, thermic Typic Kandiudults of the Orangeburg and Faceville series (Natural Resource Conservation Service (NRC) Soil Survey Geographic Database (SSURGO)). Vegetation across the 1YR and 2YR burned sites consisted of an overstory mixture of naturally regenerat ed shortleaf pine ( Pinus echinata P. Mill), loblolly pine ( P taeda L.) and longleaf pine (P palustris P. Mill) and an understory composed of annual grasses and hardwood resprouts (Clewell and Komarek, 1975; Myers and Ewel, 1990; Engstrom and Palmer, 2005 ; Glitzenstein et al., 2012) Vegetation within the unburned sites consisted of an overstory of shortleaf pine, loblolly pine and with lesser counts of

PAGE 28

28 longleaf pine and slash pine ( P elliottii Engelm.) (Clewell and Komarek, 1975) Due to the prolonged a bsence of fire, the unburned plots contained a much greater component of shade tolerant midstory and overstory hardwood species including but not limited to: sweetgum ( Liquidambar styraciflua L.), mockernut hickory ( Carya alba (L.) Nutt. ex Ell.), live oak ( Quercus virginiana P. Mill.) and water oak ( Q nigra L.) (Clewell and Komarek, 1975; Myers and Ewel, 1990) Sampling A total of nine plots were arranged in three blocks, with one representative plot per block of three prescribed fire return intervals (FR I): annual burn (1YR), biennial burn (2YR) and long unburned (40YR). To account for variability within the individual plot, each plot was comprised of nine 20 cm diameter x 10 cm height PVC sample collars (Figure 2 3) arranged in a 3 x 3 grid with 5 m sep aration following Kobziar and Stephens (2006). PVC sampling collars were constructed of Schedule 30 white 20 cm diameter pipe cut to 10 cm lengths and beveled along one edge. Collars were inserted beveled edge down into the soil or duff to a depth of app roximately 8 cm using a rubber mallet. During the course of study, any vegetative growth within the sample collars was clipped and removed prior to R s measurement. The soil collar sample sites were excluded from the annual and biennial prescribed burning in the Stoddard Plot during the months of March 2010 and March 2011. Sampling of R s 2 m 2 sec 1 ) for all plots occurred on a monthly interval over the course of two days using a LI COR Biosciences LI 8100 automated soil CO 2 sampling instrument wi th a 20 cm survey chamber (LI COR Biosciences Inc., Lincoln, NE, USA). Sampling consisted of a 120 second measurement initiated by a 15 second dead band. Concurrently with R s measurements, soil temperature (T s ) ( C) and moisture

PAGE 29

29 content (M s ) (m 3 / m 3 ) at 10 cm and 5 cm depths respectively, were recorded onboard the LI 8100 using an Omega 8831 type E T Handle temperature probe and a Decagon Systems EC 5 soil moisture probe (Omega Inc., Stamford, CT; Decagon Systems Inc., Pullman, WA), respectively. The R s study was established in the summer of 2009 with collars installed in June and July and sampling initiated in August. To account for diurnal variability, from August of 2009 until February 2010, R s T s and M s were sampled eight times per day, once per mo nth. An assessment of the preliminary results found that fewer daily measurements would sufficiently capture the diurnal variability in R s T s and M s Because of this, daily plot measurements were scaled back to three times per day (morning, mid day, an d late afternoon early evening) from March 2010 until the end of the study in May 2011. Measurements were taken on a consistent monthly interval with interruptions only due to equipment problems, heavy rain, lightning, or hazardous conditions within the p lot. The resulting dataset for the entire twenty one month study totaled 7566 R s measurements. Collected data were assessed for quality prior to analyses with strong outliers in R s T s and M s attributed to measurement or equipment error excluded from th e analysis. Recorded soil moisture content values less than 0.00, and soil temperature measurements greater than 40 C were excluded from the analyses as they resulted from equipment malfunction. Plot characteristics and vegetative sampling were assessed i n the winter and spring of 2011. Overstory vegetation was sampled using a 15 m radius circular plot (0.07 ha) centered on the middle R s sample collar. In addition to R s T s and M s the following field parameters with abbreviation and unit were recorded per R s sample collar: linear distance (m) from the sample collar to the nearest tree with a diameter at

PAGE 30

30 1.3 m height (DBH) > 10 cm (Dnearest), diameter (cm) at breast height of the nearest tree to the sample collar (DBH), mean litter depth (cm) from three measurements within 30 cm of the sample collar (Litter), mean duff depth (cm) from three measurements taken within 30 cm of the sample collar (Duff), and total mean duff and litter depth (cm) from three measurements taken within 30 cm of the sample collar (DL) (Table 2 1). The following stand condition parameters with abbreviation and units were recorded one time per sample plot: total basal area (BA) (m 2 ha 1 ), pine basal area (P BA) (m 2 ha 1 ), hardwood basal area (HW BA) (m 2 ha 1 ), and stand density (TPH ) (trees ha 1 ) (Table 2 2). In addition, the following meteorological and climatic conditions for the entire study area were recorded monthly from external sources: monthly total precipitation from the Florida Automated Weather Network Station (FAWNS) at Quincy, Florida approximately 30 km from the study site (Precip) (cm), monthly mean ambient air temperature ( C) from the FAWNS site (Temp), and monthly regional Palmer Drought Severity Index (PDSI) score from the National Oceanic and Atmospheric Administr ation (NOAA) National Climatic Data Center. Analysis Treatments were analyzed as a randomized complete block design with FRI as the main treatment. For each month, daily measurements per soil collar were averaged, and the nine soil collar means were then averaged to produce a plot level mean value for each month. This resulted in a sample size of three for each FRI treatment (total n = 9). Repeated measures analysis of variance (ANOVA) was used to test for differences in these monthly means among FRI tre atments for R s T s and M s over the twenty one monthly sampling periods between August 2009 and May 2011. Significant treatment effects were identified at p value < 0.05. To assess differences among field parameters

PAGE 31

31 by FRI, one way ANOVA tests were used Where significant differences were to assess for relationships between overall st udy period mean plot R s rates and T s M s and field parameters following Gough et al. (2004). Additional linear (Equation 2 1) and nonlinear (Equation 2 2) regression models were developed using monthly plot means per treatment and measurement season for T s M s and field parameters listed in Table 2 1. The non linear models of the relationship between R s rates and T s and M Temp were explored using an exponential equation (Equation 2 2) frequently used to describe the response of R s rates to soil temperat ure (Lundegardh, 1927; Samuelson et al., 2004; Concilio et al., 2005; Kobziar and Stephens, 2006) Following Samuelson et al. (2004) and Ryu et al. (2009) multiple regression using a forward step wise procedure was used to develop models per FRI of monthl y mean R s rates using Equation 2 3, utilizing the field and meteorological parameters that best explained the observed variability in R s rates (using R 2 and p value), while minimizing multicollinearity and BIC scores. (2 1) or (2 2) (2 3) 0 1 2 i were coefficients e stimated through regression analysis. Residuals of regressions were checked for normality and heteroscedasticity and where necessary model terms were transformed to meet assumptions.

PAGE 32

32 The 1 estimates developed using Equation 2 2 were used to estimate the Q 10 value per treatment using Equation 2 4 following Kobziar and Stephens (2006) (Lundegardh, 1927) The Q 10 value is often reported in studies of R s to describe the response of R s to a 10 C change in soil temperature (Luo and Zhou, 2006) (2 4) All statistical analyses were performed using JMP 9.0 (SAS Institute, Cary, NC, USA). Results Vegetation and groundcover among the three prescribed fire treatment types varied significantly, with the highest basal area (37.72 m 2 ha 1 ), stand density (1716.41 trees ha 1 ), duff depth (1.58 cm), and litter depth (2.81 cm) in the 40YR treatment (Table 2 2). Both the hardwood and pine components of total plot basal area increased with decreasing fire frequency (Table 2 2). Throughout the study period, across all fire return intervals, the observed soil temperature (T s ) ranged from 5.2 37.39 C (Figure 2 4). Trends in T s generally followed seasonal and monthly ambient temperature patterns with monthly mean T s in all treatments highly correlate d (R 2 = 0.81 0.90 p < 0.0001) with monthly mean 2 m temperature (M Temp) recorded at the Quincy, Florida FAWNS station. The effect of treatment on T s varied with time (treatment x time p < 0.0001), with the greatest difference between treatments general ly observed in the summer months and the least in the winter months (Table 2 3). Although soil temperature did not differ significantly among FRI treatments (p = 0.1007), the lowest temperatures were generally in the 40YR treatment and the highest general ly in the 1YR treatment (Table 2 3 and 2 4) (Figure 2 5). A distinct seasonal T s trend among the

PAGE 33

33 FRI treatments was observed, with the 1YR treatment recording the highest mean T s in the warmer spring through fall seasons and the 40YR treatment recording t he highest mean T s in the cooler winter seasons (Table 2 4) (Figure 2 4). Soil moisture content (M s ) (m 3 /m 3 ) ranged from 0 0.45 m 3 /m 3 during the study period, with M s generally highest during the winter and early spring months of the year (Table 2 5) (Fi gure 2 6). The effect of FRI on soil moisture content varied monthly (treatment x time p = 0.02), with the smallest treatment effect observed during periods of very low soil moisture (Table 2 3). Although fire return interval was not found to significant ly affect M s (p = 0.11), the lowest soil moisture content was generally observed in the 40YR treatment and the highest generally observed in the 1YR treatment (Table 2 3 and 2 5). Overall mean soil moisture content was highest in the 1YR treatment (0.17 m 3 /m 3 ), lowest in the 40YR treatment (0.12 m 3 /m 3 ), and between the two in the 2YR (0.15 m 3 /m 3 ) (Figure 2 5). Monthly mean soil moisture content was significantly negatively correlated with both monthly mean soil temperature (1YR R 2 = 0.37 p < 0.0001, 2YR R 2 = 0.32 p < 0.0001, and 40YR R 2 = 0.26 p < 0.0001) and monthly mean temperature (M Temp) (1YR R 2 = 0.40 p < 0.0001, 2YR R 2 = 0.42 p < 0.0001, and 40YR R 2 = 0.36 p < 0.0001). Starting around April of 2010, the region experienced a moderate extreme drought due to a precipitation deficit that grew in intensity until the end of the study period (Figure 2 7). The soil moisture data supported a link with the regional drought as soil moisture content had a positive linear relationship with the monthly precipita tion totals in all treatments: 1YR (R 2 = 0.17 p = 0.001), 2YR (R 2 = 0.15 p = 0.002), and 40YR (R 2 = 0.20 p = 0.0002). In addition, monthly mean M s and PDSI were moderately positively correlated: 1YR (R 2 = 0.22 p =0.0001), 2YR (R 2 =

PAGE 34

34 0.21 p = 0.0002), and 4 0YR (R 2 = 0.29 p < 0.0001). The monthly precipitation data were also only moderately linearly correlated with the regional monthly PDSI values (R 2 = 0.34 p < 0.0001). Soil CO 2 efflux rates (R s ) ranged from 0 11.98 mol CO 2 m 2 sec 1 during the study per iod, with the highest R s rates during the warmer months and the lowest rates during the cooler months (Table 2 6) (Figure 2 8). R s rates varied significantly among treatments (p = 0.0007), with the highest R s rates typically in the 40YR treatment and the lowest typically in the 1YR treatment (Table 2 3 and 2 6). The average overall 40YR mean R s rate 4.28 mol CO 2 m 2 sec 1 ) was 37 % higher than the 1YR mean rate (2.68 mol CO 2 m 2 sec 1 ) and 25 % higher than the 2YR rate (3.20 mol CO 2 m 2 sec 1 ) (Figure 2 5). The treatment effect of FRI varied monthly (treatment x time p < 0.0001), with the greatest differ ence between treatments observed during the summer months and the least during the winter months (Table 2 3). When treatments were ignored and monthly mean plot values pooled and analyzed as a group, relationships between R s T s M s and plot vegetative an d R s and T s (0.68) and R s and Mean Temp (0.77) (Table 2 7). R s also exhibited a surprisingly negative relationship with M s ( 0.34) while exhibiting a weak positive correlation with monthly precipitation (0.11) (Table 2 7). In the same test R s also exhibited correlations with the following plot level vegetative characteristics: basal area (0.34), stand density (0.33), distance to nearest tree ( 0.31), duff depth (0.36), litter depth (0.29) and total duff+litter depth (0.36). Soil temperature was only weakly

PAGE 35

35 correlated with the same plot level vegetative characteristics (Table 2 7) ind icating that while T s explained much of the temporal variation in R s it did not explain the differences in R s observed between the treatment types. In the linear regressions, overall mean R s rates were significantly positively correlated with stand basal area (R 2 = 0.75 p = 0.003), and stand density (R 2 = 0.76 p = 0.002) and significantly negatively correlated with the distance to the nearest tree (R 2 = 0.68 p = 0.006), although the diameter of the nearest tree was not significant (Figure 2 9). In additi on, litter depth (R 2 = 0.64 p = 0.001), duff depth (R 2 = 0.87 p = 0.0003), and litter+duff depth (R 2 = 0.91 p < 0.0001) were significantly correlated with overall mean R s rates (Figure 2 10). To assess the influence of vegetative and meteorological paramet ers on monthly mean R s rates within treatments, simple linear regression models (Equation 2 1) were developed for each parameter and FRI (Table 2 8). Linear regression indicated that a strong positive relationship between monthly mean R s rates and T s exis ted for the 1YR (R 2 = 0.62, p < 0.0001), 2YR (R 2 = 0.78, p < 0.0001), and 40YR (R 2 = 0.65, p < 0.0001) fire return intervals (Table 2 8) (Figure 2 11). Nonlinear exponential models (Equation 2 2) used to explore the relationship between monthly mean R s an d T s by treatment reported similar fit to linear models (Table 2 9) (Figure 2 12) (Lundegardh, 1927) 0 1 from the nonlinear models were similar to estimates reported by Samuelson et al. (2004) and Kobziar and Stephens (2006). Similar to T s monthly mean ambient air temperature (M Temp) as recorded by the FAWNS station, also ha d a strong positive linear (R 2 = 0.61 0.77) and nonlinear relationship with R s (R 2 = 0.63 and 0.79, respectively) (Figure 2 13) (Figure 2 14). Soil moisture content (M s ) had a very weak negative linear relationship with monthly mean R s for the 1YR (R 2 =

PAGE 36

36 0.14, p = 0.004), 2YR (R 2 = 0.14, p = 0.003), and 40YR (R 2 = 0.03, p = 0.22) treatments (Table 2 8) (Figure 2 15). Monthly total precipitation did not have a notable (R 2 > 0.10) relationship with monthly mean R s For the following variables analyzed with in each treatment there was either no significant simple linear regression relationship with monthly mean R s or the relationship was R 2 < 0.10: distance to nearest tree, diameter at breast height of the nearest tree, litter depth, duff depth, total duff +l itter depth, plot basal area, pine species plot basal area, hardwood species plot basal area, plot tree density, or monthly total precipitation. While the plot level variables did not correlate with monthly mean R s rates, they did vary significantly among the treatments (Table 2 2). To assess seasonal variations in the relationship between R s and T s and R s and M s simple linear (T s and M s ) and nonlinear (T s ) regression models were developed per treatment and season (fall, winter, spring, and summer) (Table 2 10, 2 11, and 2 12). In the T s linear models, for all treatments, monthly mean R s was most closely correlated with T s during the fall and winter months (R 2 = 0.69 0.90) and weakly correlated during the spring and summer months (R 2 = 0.11 0.63). Si milarly in the nonlinear T s models, for all treatments, monthly mean R s was most closely correlated with T s during the fall and winter months (R 2 = 0.66 0.96) and least correlated during the spring and summer months (R 2 = 0.10 0.62). In the soil moist ure content linear models, monthly mean R s was most closely correlated with M s during the summer and fall seasons (R 2 = 0.13 0.82), but the relationship was only significant (p < 0.05) in the 40YR treatment. To assess the drivers of R s and to determine i f the drivers varied per treatment, all variables (Table 2 1) were tested for their effect on monthly means of R s for each

PAGE 37

37 treatment in three separate forward step wise multiple linear regression procedures (Equation 2 4) (Table 2 13). In each of the deve loped models, either T s or M Temp were the first terms added, explaining > 50% of the variability of R s Soil M s was the second term added for all treatments, which was then followed by distance to the nearest tree (D Nearest) in the 1YR and 40YR models, and pine basal areas (P BA) in the 2YR model. A negative correlation with D Nearest was observed in both the 1YR and 40YR models. Overall the multiple linear regression models by treatment fit the data (R 2 = 0.82 0.89) only slightly more than the best fi tting linear and nonlinear regression models. The 2 were used in the Q 10 model (Equation 2 4) to describe the incremental response of R s to a change of 10 degrees C in soil temperature (Table 2 9) (Table 2 11) (Lundegardh, 1927) The annual Q 10 values (1YR = 1.65, 2YR = 1.96, and 40YR = 2.16) were similar to those reported by Kobziar and Stephens (2006) for a Sierra Nevada pine plantation and Xu et al. (2011) for an oak forest in Missouri, USA, but lower than those reported by Maier and Kress (2000) for a loblolly pine p lantation in North Carolina, USA. Total monthly and annual soil carbon emissions per prescribed fire interval were estimated using the linear regression models of R s responses to changes in ambient temperature (Table 2 6) following Samuelson et al. (2004). Twenty four hour 2 m elevation ambient temperature measurements recorded hourly from August 1, 2009 July 31, 2010 at the Quincy, FL FAWNS site located approximately 30 km from the study sites and offering the most consistent data available were used as i nputs to predict hourly R s rates. The predicted R s rates ( mol CO 2 m 2 sec 1 ) were then

PAGE 38

38 converted to hourly soil carbon fluxes (g C m 2 hr 1 ) which were then summed to estimate monthly and annual soil carbon fluxes. Estimated total monthly soil carbon emissions (Figure 2 16) were consistently higher in the unbur ned 40YR treatment than the frequently burned 1YR and 2YR treatments. Similarly, estimated total annual soil carbon emissions per treatment showed the highest soil carbon efflux in the 40YR treatment (1688 g m 2 y 1 ) and the lowest in the 1YR (1069 g m 2 y 1 ) and 2YR (1268 g m 2 y 1 ) treatments. Discussion The plot level monthly mean R s rates observed in our study ( 0.56 9.16 mol CO 2 m 2 sec 1 ) were similar on average, but ranged much higher than those reported in a Georgia, USA loblolly pine plantation (1.27 5.59 mol CO 2 m 2 sec 1 ) (Samuelson et al., 2004) a North Carolina, USA loblolly pine plantation (0.5 6.0 mol CO 2 m 2 sec 1 ) (Maier and Kress, 2000) and a Sierra Nevada, California, USA CA ponderosa pine Jeffrey pine plantation (2.37 4.55 mol CO 2 m 2 sec 1 ) (Kobziar and Stephens, 2006), although each of those particular study locations had management tenures, s oils, climates, and ecosystems different from those investigated here The higher range of R s rates observed in our sites relative to those mentioned previously were likely related to the length of the growing seasons, warm annual temperatu res, annual precipitation, and high stand biomass associated with the long unburned treatment at the Tall Timbers Research Station which may have driven higher autotrophic and heterotrophic respiration rates The results of our study indicate that the old field forest conditions that were shaped by over 50 years of frequent prescribed fire (or a lack of burning during the same period) can cause significant differences in overall mean and monthly mean soil CO 2 efflux rates.

PAGE 39

39 Across all treatments on a monthly basis, soil temperature and monthly mean 2 m temperature (M Temp) explained more of the temporal variability of R s rates than any other recorded vegetative or meteorological parameter. In our simple linear and nonlinear regression models of monthly mean R s rates, T s and M Temp explained the majority of the variation in R s compared to the other parameters and in the forward step wise multiple linear regression procedures T s or M Temp were the first terms to be entered into any of the models. These results are consistent with others who have investigated the drivers of R s rates in S outheastern forest systems and found strong correlations with soil temperature (Fang et al., 1998; Gough and Seiler, 2004; Samuelson et al., 200 4, 2009; Gough et al., 2005) The R s correlation with T s reported in this study (R 2 = 0.62 0.78 in linear models and R 2 = 0.60 0.80 in nonlinear models) was higher than those reported (R 2 = 0.38 0.56) by Samuelson et al. (2004) for a Georgia, USA lo blolly pine plantation, much higher than those (R 2 = 0.26) reported by Gough and Seiler (2004) for a loblolly pine plantation in South Carolina, USA, and similar to those (R 2 = 0.70) reported for a loblolly pine plantation in North Carolina, USA by Maier a nd Kress (2000). While T s explained the majority of the temporal variability in R s rates, a lack of significant differences for T s among treatments suggests that other factors such as vegetation explained the differences in R s rates between the treatments Soil moisture content (M s ) also explained some of the temporal variability in R s rates even though significant differences in soil moisture content were not observed among s was moderately negatively correlated with R s ( 0.34) when treatments were ignored, and in the treatment specific regression models a weak negative linear relationship was in the

PAGE 40

40 1YR, 2YR, and 40YR treatments (R 2 = 0.14, 0.14, and 0.03, respectively). In contrast, the forward step wise multiple linear regression models by treatment identified a positive relationship with M s as the second most significant term behind T s or M Temp in explaining monthly mean R s rates. The fact that soil moisture content explained some but relatively l ittle of the variability in R s rates is consistent with the results of other studies of R s in the S outheast as soil moisture is rarely limited in these systems (Fang et al., 1998; Gough and Seiler, 2004; Samuelson et al., 2009). The positive and negative relationships identified between R s and M s may have been the result of interactions between temporal patterns of precipitation, seasonal plant soil water use, seasonal changes in R s rates due to temperature and plant growth patterns, and the effects of dro ught on soil water content and vegetation. The widespread regional drought during much of our study likely resulted in near term reduced plant photosynthetic activity and belowground carbon allocation and root respiration (R a ), as well as reduced heterotr ophic microbial activity and respiration due to limited moisture availability (R h ) (Ryan and Law, 2005; Cisneros Dozal et al., 2007). This is supported by the results of a recent multi year eddy covariance study of north Florida slash pine plantation carb on dynamics which revealed that during periods of drought stress above ground carbon assimilation and total ecosystem respiration, including R s were reduced relative to non drought periods (Bracho et al., 2012). Future and longer term studies of R s rates in this system during drought and non drought periods may better elucidate the impacts of prescribed fire management regime on the connectivity of soil water content, R s rates, and forest carbon assimilation.

PAGE 41

41 While soil temperature explained the majority of the temporal variation in R s rates in our study, it did not explain the observed differences among the prescribed fire treatments. we suggest that the observed differences in mean R s rates among treatments were related to the amount of total abovegro und carbon in living biomass in the treatments (1YR 50 tons ha 1 2YR tons ha 1 and 40YR tons ha 1 ) (Kevin Robertson unpublished data.). In support of this, previous research has found no significant differences in soil carbon, nitrogen, or phosphorus content among treatments suggesting that soil carbon and nutrient content were not responsible for the observed differences in R s between treatments (Kevin Robertson unpublished data). Other studies investigating R s rates across stand age and biomass gradients have found mean R s rates to be high er in older stands with greater aboveground biomass (Ewel et al., 1987a; Amiro et al., 2010). In a trenching and exclusion experiment along a chronosequence of temperate forests in China, Luan et al. (2011) found that R s rates were significantly (R 2 = 0.5 9 p < 0.05) correlated with site basal area. While our research did not investigate belowground living root biomass among the treatments we suspect that total belowground biomass would follow similar trends among the treatments as aboveground biomass. P revious research has shown that R s rates tend to be positively correlated with root biomass (Lou and Zhou, 2006). In our study both 7) and linear regression identified relationships between overall mean R s rates and plot ba sal area (R 2 = 0.75 p = 0.003) and stand density (R 2 = 0.76 p = 0.002) when treatments were ignored and the data pooled. In a review of the controls and correlates of R s across multiple ecosystems Raich and Tufekciogul (2000) found evidence supporting pos itive correlations between R s rates and aboveground

PAGE 42

42 productivity in grasslands (R 2 = 0.80 p < 0.01) and R s rates and litterfall production in forests (R 2 = 0.90 p < 0.001). In this study we suggest that the weak within treatment relationships between R s rates and plot basal area and stand density were due to the relatively low range of within treatment variability of those characteristics. The differences in R s rates among treatment types in our study may have also been associated with variations in fore st composition, as the long unburned sites had a much greater component of deciduous tree species than the frequently burned sites. Soil CO 2 efflux rates have been shown in other studies to differ between forest types, with lower rates often reported in c oniferous forests than in broad leafed forests of the same soil type (Raich and Tufekciogul, 2000) In a study of R s rates in a mixed coniferous deciduous forest in Belgium, Yuste et al. (2005) found that mean R s rates were lower under conifer tree cano pies than under deciduous canopies, with total estimated annual carbon flux approximately 50% greater in the deciduous sites (8.8 2.2 Mg C ha 1 yr 1 ) than in the coniferous sites (4.8 0.7 Mg C ha 1 yr 1 ). In their review Raich and Tufekciogul (2000) s uggested that the observed differences in R s between coniferous and deciduous forests may have been driven by forest litter production and quality, carbon allocation, and autotrophic contributions to total R s It is important to note however, in nearly al l of the cases mentioned previously experimental partitioning of the relative contributions of autotrophic and heterotrophic sources of R s had not been assessed. Litter has been shown to be a significant source of labile carbon for heterotrophic respiratio n (Sayer, 2006). In a study of a chronosequence of deciduous forests in China, Luan et al. (2011) suggested that labile carbon quantity and quality from leaf

PAGE 43

43 litter primarily drove the observed spatial variation in R h rates. Furthermore, experimental man ipulations in litter addition and exclusion have shown to both positively and negatively (respectively) influence R h rates in multiple ecosystems (Bowden et al., 1993; Sayer, 2006; Chemidlin Prevost Boure et al., 2010; Sulzman et al., 2012). In our study the ratio of hardwood basal area to pine basal area (Table 2 2), litter and duff depths (Table 2 2), and mean annual litter and duff loads (1YR = 4.62, 2YR = 5.42, and 40YR = 5.47 t ha 1 yr 1 respectively) were highest in the 40YR treatment, suggesting th at those factors contributed to the differences in R s rates among treatments. Though not quantified in our study, observations in the plots suggest that the litter in the 40YR treatment and lesser so in the 2YR treatment was dominated by deciduous leaf li tter with a mixture of pine needles and few understory forbs and grasses. On the other hand, similar to the findings of Robertson and Ostertag (2007), observations in the 1YR treatments suggested that the leaf litter was dominated by pine needles, and ann ual understory forbs and grasses, and contained relatively little deciduous leaf litter. Differences in the composition of the leaf litter by treatment type may have been associated with variations in the quality of the leaf litter with regards to the lit ter as microbial substrate. In our study we did not assess the carbon and nutrient content of the litter samples, however in an old field loblolly shortleaf pine successional study, Hinesley and Nelson (1991) reported that the quality of the litter increa sed with seral stage, as N, P, K, Ca, and Mg increased 15%, 20%, 75%, 202%, and 72% respectively, between early sere mature pine forests and late sere mixed hardwood forests. Given this, the observed differences in R s rates between treatments may have bee n due to

PAGE 44

44 variations in the relative contributions of R h caused by litter composition, quality, depth, and loading. The results of our study also suggest that forest structure may have had some influence on the spatial variability of R s rates, likely throug h the variable contributions of root and mycorrhizal respiration to R s In our within treatment multiple regression models, in addition to temperature and soil moisture, R s rates were negatively associated with distance to nearest tree in the 1YR and 40YR treatments and positively associated with pine basal area in the 2YR treatment. In addition, in the overall simple linear regression models when treatments were ignored and all plots grouped, R s rates were significantly negatively correlated with distanc e to the nearest tree (R 2 = 0.68 p = 0.006) but not at all correlated with the diameter of the nearest tree. These results contrast with the findings of a study in a tropical rainforest in Borneo that found R s to be highly correlated (R 2 = 0.60 p < 0.001) with the diameter of trees within 6 m of the sampling point (Katayama et al., 2009) It is likely that the difference between our results and those of the Borneo study may have been due to biome related differences in forest composition and species, as K atayama et al. attributed much of the spatial variation in R s to the influence of very large (DBH > 60 cm) Dipterocarpaceae trees. A similar study by Samuelson et al. (2004) found a small but significant difference in R s rates based on the horizontal samp ling position within a loblolly pine plantation, with higher R s rates closer to the base of trees (Samuelson et al., 2004) This is not surprising given the very spatially explicit linear arrangement of pine plantations which contrast greatly to the natur ally regenerated stands at Tall Timbers where the arrangement of trees and their root growth is more variable Further study explicitly

PAGE 45

45 designed to address the linear effect of distance to nearest trees on R s rates (similar to Clinton et al. (2011)) may i dentify statistically stronger effects than we were able to identify in our study design Prescribed fire may have affected the heterotrophic soil microbial populations and thus R h rates in the 1YR and 2YR treatments. Given that the annual litterfall ra tes were similar among treatments, while litter and duff depths differed significantly among treatments (Table 2 2), we attribute much of the difference in litter and duff depths to the direct combustion of surface fuels by the fast moving, l ow intensity prescribed fires in the frequently burned sites. The combustion of soil surface litter and duff material could result in an immediate reduction in forest floor and duff inhabiting heterotrophic organisms through direct consumption and tempera ture related mortality (DeBano, 1998, Neary, 1999, Choromanska and DeLuca, 2002; Certini, 2005). While a temporary post fire decline in surface dwelling heterotrophic populations is expected as fires of this type have been shown to increase soil temperatu res at 5 cm depth to near 150C we suspect that prescribed fire resulted in little direct temperature related mortality of the deeper subsurface soil microbial communities (Debano, 2000; Certini, 2005) Following fire, depending on fire intensity and soil characteristics, there is generally both a direct loss of surface organic material due to combustion and also a deposition of organic carbon, nitrogen, and phosphorus in the form of ash and parti ally combusted material (Neary, 1999; Choromanska and DeLuca, 2002; Certini, 2005). The loss of surface litter and duff can reduce labile carbon available for microbial decomposition, while the post fire pulse of available carbon, nitrogen, and phosphorus can result in increased soil microbial activity in the post fire period due to the release of

PAGE 46

46 previously bound nutrients (Neary, 1999; Certini, 2005). As no sample collars were designated as unburned controls in the 1YR and 2YR treatments, we were unable to test for the specific effect of an individual prescribed burn on monthly R s rates. The results of an experiment in a frequently burned loblolly pine forest in South Carolina, USA, however found no significant impact on forest soil microbial enzyme act ivity following a low intensity prescribed fire (Boerner et al., 2006). The estimated Q 10 values that describe the response of R s to a change in temperature, suggested that the relative contributions of heterotrophic (R h ) and autotrophic (R a ) respiration to total R s may have existed among the treatment types. The annual Q 10 values in our study 1.65 (1YR), 1.96 (2YR), and 2.16 (40YR) suggest that higher contributions of soil microbial sources (R h ) drove higher R s rates in the 40 YR treatment and lesser so in the 2YR and 1YR treatments. This is because others have reported that the Q 10 temperature response of autotrophic and heterotrophic sources of R s differ, with heterotrophic soil microorganisms more sensitive to changes in soi l temperature than plant associated autotrophic sources (Bhupinderpal Singh et al., 2003; Zhou and Zhou, 2012) Bhupinderpal Singh et al. (2003) reported following a girdling experiment in a boreal Scots pine forest, that R s from heterotrophic sources dro pped significantly following a 6 C decline in soil temperature, while autotrophic sources of R s were unchanged during that same period. Similarly, Zhou and Zhou (2012) reported in a review of several studies that average Q 10 values were less for roots (2. 07) than litter (2.68) or bulk soil organic matter (2.54). Likewise, in a trenching and exclusion experiment along a chronosequence of temperate forest sites in China,

PAGE 47

47 Luan et al. (2011) reported that Q 10 values for heterotrophic sources of R s were higher than for autotrophic sources, regardless of stand age. We suggest however that using Q 10 values as a proxy for experimental partitioning methods should be considered with caution, as our results and those mentioned previously, contrast with others t hat have found Q 10 values for plant roots to be higher than the heterotrophic R s sources in the bulk soil (Boone et al., 1998; Saiz et al., 2006) The relative partitioning of R s suggested by our Q 10 values also contradicts the results of a review of glob al R s studies that found the ratio of R h contributions to total R s declined with both increasing R s and ecosystem productivity (Subke et al., 2006). Both Subke et al. (2006) and Kuzyakov and Gavrichkova (2010) suggested that the proportional decline in R h contributions to total R s might be due to an increase in total belowground productivity across the treatments in our study, the increase in total R s basal area, stand densi ty, and estimated total aboveground biomass in the 40YR sites relative to the 1YR and 2YR sites, may have led to higher R a contributions relative to R h in the 40YR sites, contrasting the results of our Q 10 values. In our study it was observed that the relationship between R s and soil temperature (T s ) and soil moisture content (M s ) varied seasonally. We suggest that the changes were indicative of phenological shifts in the relative contributions of R a and R h to R s Previous research in partitioning stud ies has shown that during periods of aboveground vegetative growth, R a contributions to R s can increase relative to R h as plants allocate recent C photosynthate belowground, driving higher root maintenance, root growth, and mycorrihizal fungal respiration rates (Subke et al., 2006; Kuzyakov and Gavrichkova,

PAGE 48

48 2010). Consequently, it has been shown that during periods of aboveground vegetative growth, the T s and R s relationship weakens as other variables such as soil moisture (M s ) and available photosyntheti cally active radiation become more important in governing belowground C allocation by plants (Ekblad and Hogberg, 2001; Davidson et al., 2006; Wertin and Teskey, 2008) Fenn et al. (2010) reported similar results for a multi season study of R s rates in a woodland in Oxfordshire, UK, with soil temperature explaining less of the variation in R s during the summer months than during the spring. Our data tend to support these previous observations as the seasonal relationships between R s and T s were strongest during the fall and winter seasons in all treatments and the seasonal relationship between R s and M s were strongest during the summer season in the 2YR and 40YR treatments. While spatial temporal correlations between T s and R s have been well documented in the literature for many ecosystems, soil temperature has also been identified as a potentially confounding variable in R s studies by masking other phenological and meteorological variables that more directly govern the physiological mechanisms of autotrop hic soil CO 2 production (Hogberg et al., 2009). It was interesting to observe the seasonal reverse of soil temperature trends between the 1YR and 40YR treatments wherein the warmest soils were found in the summer in the 1YR plots and in the winter in the 4 0YR plots. These temperature variations were likely the result of the differences in canopy cover between the treatments and the effect of canopy cover on the balance of incoming shortwave radiation and outgoing longwave radiation These seasonal soil te mperature fluctuations were similar to results reported by Samuelson et al. (2004) in a study of loblolly pine plantation management types. In their study, instead of prescribed fire,

PAGE 49

49 herbicidal weed control resulted in a similar seasonal soil temperature reversal (Samuelson et al., 2004) Herbicidal weed control in some cases can alter forest structure similar to frequent low intensity prescribed fire, suggesting that long term management induced changes in vegetative cover may have caused the seasonal t emperature shift among treatment types (Brose and Wade, 2002) A linear transect study of forest cover effects on soil temperature from New York, USA, found that forest cover increased soil temperature in the winter and decreased it in the warmer months r elative to an open field (Michelsen Correa and Scull, 2005) Even though a positive correlation was observed between R s and T s a similar seasonal change in the hierarchal order of R s rates among treatments was not observed in our study. This is comparab le to a disconnect between soil temperature and R s following prescribed burning reported by Ryu et al. (2009) in a mixed conifer forest in California, USA. In their study, the authors reported that prescribed fire reduced R s while simultaneously increasin g soil temperature and moisture content (Ryu et al., 2009) The seasonal disengagement between R s rates and T s observed in our study further suggest that mechanisms other than soil temperatures were driving soil CO 2 efflux rates. The R s rates observed in this study resulted in reduced estimated total annual soil carbon emissions in the 1YR (37%) and 2YR (25%) treatments relative to the long unburned stands. The estimated monthly carbon emissions for all treatments were similar to those reported by Samuels on et al. (2004) for an unburned loblolly pine plantation in southwestern Georgia, USA, while the estimated total annual soil carbon emissions reported in our study, 1069 g m 2 y 1 (1YR), 1268 g m 2 y 1 (2YR), and 1688 g m 2 y 1 (40YR), were similar to tho se (1410 g m 2 y 1 ) reported by Maier and Kress

PAGE 50

50 (2000) for an unburned loblolly pine plantation, but less than those (778 966 g m 2 y 1 ) reported by the Samuelson et al. (2004) study. Conclusion A forest management regime employing frequent prescribed fire can fundamentally alter forest structure and composition relative to a fire exclusion regime resulting in reduced soil carbon fluxes in burned vs. fire excluded trea tments It was reported that average R s rates in the annually burned forests were approximately 37% lower than those of the long unburned forests, while total estimated annual soil carbon fluxes were also lower in the annually (1069 g m 2 y 1 ) and biennia lly (1268 g m 2 y 1 ) burned forests than in the long unburned forest (1688 g m 2 y 1 ). Our results indicate that the differences in R s rates between treatments were driven by variations in the amount and composition of forest vegetation and litter and duf f depths between treatments sites. Similar to the results of others (Raich and Tufekciogul, 2000) we found that mean R s rates were higher at sites with greater aboveground biomass and litter and duff depths. We suggest that these conditions were responsi ble for the variations in R s rates due to increased total belowground carbon allocation by plants and labile carbon for heterotrophic soil microbes, respectively. To assess the full effect of prescribed fire management on total ecosystem carbon dynamics, f uture research is needed that also quantifies aboveground carbon gains and losses including losses due to combustion. Additional research is also needed to understand the implications of prescribed fire season on R s rates, as some forest lands managed for conservation or ecosystem restoration in the region apply prescribed fire during the growing season while still others burn during the dormant season. Given that the season in which prescribed fire is applied can have significant impacts on forest

PAGE 51

51 specie s composition and structure, it is not known whether corresponding changes in R s rates occur (Waldrop et al., 1992). Regardless of prescribed fire management regime and consistent with previous studies, soil temperature explained well over half of the var iability in R s rates in old field forests. Interestingly, the strength of temperature models of R s rates varied seasonally, with the least predictive power occurring in models of the warmer spring and summer months during the growing season. Future effor ts to model carbon dynamics under elevated temperatures should address this seasonal variability in the relationship between soil temperature and R s rates. Furthermore, research is needed to understand the specific response of autotrophic and heterotrophi c sources of R s to changes in factors other than temperature, such as aboveground litter inputs and forest management practices.

PAGE 52

52 Table 2 2 efflux rates at the Tall Timbe rs Research Station, FL Parameter category Plot variable Abbreviation Measured Measurement location Microclimate Soil temperature T s (C) 3x daily 5 15 cm of collar Soil moisture content M s (m 3 /m 3 ) 3x daily 5 15 cm of collar Vegetation Basal area B A (m 2 ha 1 ) Winter 2011 15 m radius circular plot from center collar Pine basal area PBA (m 2 ha 1 ) Winter 2011 15 m radius circular plot from center collar Hardwood basal area HWBA (m 2 ha 1 ) Winter 2011 15 m radius circular plot from center collar Stand density TPH (trees ha 1 ) Winter 2011 15 m radius circular plot from center collar Distance to the nearest tree Dnearest (m) Spring 2011 Linear distance from soil collar to nearest tree (DBH > 10 cm) Diameter of the nearest tree DBH (cm) Spring 2011 DBH of the nearest tree measured in Dnearest Forest floor Duff depth Duff (cm) Spring 2011 Avg. of three measurements within 30 cm of collar Litter depth Litter (cm) Spring 2011 Avg. of three measurements within 30 cm of collar Total duff and li tter depth DL (cm) Spring 2011 Avg. of three measurements within 30 cm of collar Weather Total precipitation Precip (cm) Monthly Quincy, FL FAWNS station Mean air temperature (2 m) Temp (C) Monthly Quincy, FL FAWNS station Palmer drought severity ind ex PDSI Monthly Northwest Florida regional estimate from NOAA NCDC

PAGE 53

53 Table 2 2. Mean forest characteristics per prescribed fire treatment type at the Tall Timbers Research Station, FL FRI Year Trees (ha) Hardwoo d basal area (m 2 ha 1 ) Pine basal area (m 2 ha 1 ) Total basal area (m 2 ha 1 ) Mean d iameter at breast height (DBH) (cm) Distance to nearest tree (m) DBH of nearest tree (cm) Duff depth (cm) Litter depth (cm) Annual litterfall (t ha 1 yr 1 ) 1YR 2011 282.93 (64.83) b 3.87 (6.28) b 7.92 (3.68) a 1 1.79 (7.22) b 16.48 (5.50) a 3.26 (0.90) a 10.38 (3.81) c 0.08 (0.06) c 1.77 (0.91) c 4.63 (1.53) a 2YR 2011 400.81 (344.87) b 6.30 (4.28) ab 9.16 (6.14) a 15.45 (2.15) b 22.26 (8.14) a 3.18 (1.40) a 27.73 (7.82) a 0.46 (0.41) b 2. 17 (0.79) b 5.42 (0.69) a 40YR 2011 1716.41 (681.42) a 15.73 (3.59) a 21.99 (11.22) a 37.72 (8.36) a 12.77 (5.82) a 1.48 (0.16) b 14.22 (2.40) b 1.58 (0.55) a 2.81 (0.58) a 5.47 (1.37) a Data presented are means of three sample plots per FRI treatment. Data in parentheses are standard deviation. FRI is 1953). Distance to nearest tree is the average distance from each soil respiration sampling point to the nearest tree (DBH > 10 cm). DBH of the nearest tree is the average DBH of the nearest tree to each soil respiration sampling point. Litter and duff depth are the average of 18 measurements taken one time per sample plot. Litterfall rates provided by K. Robertson (unpublished data).

PAGE 54

54 Table 2 3. Results of the repeated measures ANOVA for soil CO 2 efflux (R s ), soil temperature (T s ), and soil moisture content (M s ) means at the Tall Timbers Research Station, FL R s T s M s Analysis period Source df F P > F df F P > F df F P > F Total Month 20 105.19 < 0.0001 18 306.76 < 0.0001 19 97.51 < 0.0001 Treatment*Month 40 3.25 < 0.0001 36 4.04 < 0.0001 38 1.68 0.0190 Treatment 2 20.72 0.0007 2 3.0 9 0.1007 2 2.95 0.1130 Fall Month 5 114.16 < 0.0001 5 241.78 < 0.0001 5 66.73 < 0.0001 Treatment*Month 10 4.12 0.0012 10 1.31 0.2684 10 2.06 0.0611 Treatment 2 17.65 0.0031 2 2.70 0.1457 2 4.26 0.0705 Winter Mon th 5 25.16 < 0.0001 3 61.75 < 0.0001 5 25.45 < 0.0001 Treatment*Month 10 1.96 0.0761 6 0.40 0.8650 10 1.96 0.0753 Treatment 2 13.40 0.0061 2 10.48 0.0036 2 6.30 0.0335 Spring Month 5 25.61 < 0.0001 5 243.07 < 0.0001 5 75. 96 < 0.0001 Treatment*Month 10 2.73 0.0168 10 5.50 0.0001 10 1.03 0.4454 Treatment 2 11.24 0.0054 2 4.11 0.0646 2 1.41 0.3003 Summer Month 2 16.70 0.0003 2 38.05 < 0.0001 1 139.44 < 0.0001 Treatment*Month 4 3.78 0.032 7 4 4.70 0.0163 2 3.02 0.1235 Treatment 2 16.03 0.0039 2 9.91 0.0126 2 1.32 0.3340 For each month, daily measurements per soil collar were averaged, and the nine soil collar means were then averaged to produce a plot level mean value for each month. This resulted in a sample size of three for each FRI treatment (total n=9). The effect of month, treatment, and treatment*month on plot level means were tested for significance (p < 0.05).

PAGE 55

55 Table 2 4. Mean seasonal and total study period soil temperature per prescribed fire treatment type at the Tall Timbers Research Station, FL FRI Fall mean T s (C) Winter mean soil T s (C) Spring mean T s (C) Summer mean T s (C) Study mean T s (C) 1YR 21.74 (4.79) a 11.86 (3.81) b 22.04 (6.21) a 28.84 ( 2.90) a 20.45 (7.16) a 2YR 21.44 (4.35) a 12.20 (3.59) ab 20.67 (4.74) a 27.81 (1.64) ab 19.98 (6.37) a 40YR 21.10 (3.50) a 12.93 (3.13) a 18.90 (3.26) a 25.90 (0.82) b 19.42 (5.19) a Values are means with standard deviations in parentheses. Letters pe r column show significant differences between fire

PAGE 56

56 Table 2 5. Mean seasonal and total study period soil moisture content per prescribed fire treatment type at the Tall Timbers Research Station, FL FRI Fal l mean M s (m 3 /m 3 ) Winter mean M s (m 3 /m 3 ) Spring mean M s (m 3 /m 3 ) Summer mean M s (m 3 /m 3 ) Study mean M s (m 3 /m 3 ) 1YR 0.16 (0.06) a 0.25 (0.06) a 0.15 (0.09) a 0.15 (0.09) a 0.20 (0.08) a 2YR 0.14 (0.07) a 0.22 (0.08) ab 0.14 (0.09) a 0.12 (0.09) a 0.17 (0.09 ) a 40YR 0.11 (0.06) a 0.20 (0.08) b 0.12 (0.08) a 0.10 (0.07) a 0.15 (0.09) a Values are means with standard deviations in parentheses. Letters per column show significant differences between fire Table 2 6. Mean seasonal and total study period soil CO 2 efflux per prescribed fire treatment type at the Tall Timbers Research Station, FL FRI Fall mean R s 2 m 2 sec 1 ) Winter mean R s 2 m 2 sec 1 ) Spring mean R s 2 m 2 sec 1 ) Summer mean R s 2 m 2 sec 1 ) Study mean R s 2 m 2 sec 1 ) 1YR 3.36 (1.71) b 1.24 (0.87) b 2.56 (1.21) b 4.93 (1.95) b 2.67 (1.87) b 2YR 3.75 (2.02) b 1.60 (1.24) b 3.09 (1.39) b 5.49 (2.11) b 3.09 (2.12) b 40YR 4.93 (2.36) a 2.61 (1.49) a 3.98 (1.94) a 7.59 (2.25) a 4.22 (2.52) a Values are means with standard deviations in parentheses. Letters per column show significant differences between fire

PAGE 57

57 Table 2 2 efflux (R s ), soil temperature (T s ), soil moisture content (M s ) and plot vegetative and meteorological characteristics Variable R s 2 m 2 sec 1 ) T s (C) M s (m 3 /m 3 ) R s 2 m 2 sec 1 ) 1.00 0.68 0.34 T s (C) 0.68 1.00 0.53 M s (m 3 /m 3 ) 0.34 0.53 1.00 Dist nearest (m) 0.31 0.10 0.09 DBH nearest (cm) 0.01 0.00 0.07 Stand density (tree ha 1 ) 0.33 0.09 0.17 Basal area (m 2 ha 1 ) 0.34 0.12 0.23 Hardwood basal area (m 2 ha 1 ) 0.30 0.12 0.19 Pine basal area (m 2 ha 1 ) 0.26 0.08 0.18 Duff depth (cm) 0.36 0.10 0.23 Litter depth (cm) 0.29 0.12 0.14 Duff+litter depth (cm) 0.36 0.12 0.20 Monthl y temp ( C) 0.77 0.91 0.61 Monthly precip (cm) 0.11 0.03 0.40 Correlations are of monthly mean plot measurements with treatments ignored and all treatment x plot x month means pooled. R s is soil CO 2 2 m 2 sec 1 ), T s is soil temperature (C), M s is soil volumetric moisture content (m 3 /m 3 ). Plot vegetative and meteorological characteristics described further in Table 2 1.

PAGE 58

58 Table 2 8. Linear regression relationships between soil CO 2 efflux rates and field conditions by fire return interval FRI Variable Model and estimates R 2 p 1YR T s R s = 0.5101 + 0.16029* T s 0.62 < 0.001 1YR M s R s = 3.6182 6.3661* M s 0.14 0.004 1YR Mean Temp R s = 0.4641 + 0.1799*M Temp 0.76 < 0.001 2Y R T s R s = 1.1736 + 0.2218* T s 0.78 < 0.001 2YR M s R s = 4.1907 7.6204* M s 0.14 0.003 2YR Mean Temp R s = 0.1556 + 0.1918*M Temp 0.77 < 0.001 40YR T s R s = 2.0303 + 0.3337 T s 0.65 < 0.001 40YR M s R s = 4.6763 4.6868* M s 0.03 0.215 40YR Mean Te mp R s = 0.2708 + 0.2291*M Temp 0.61 < 0.001 Model data are mean monthly measurements from August 2009 May 2011 taken at the Tall Timbers Research Station near Tallahassee, Florida, USA. R s is soil CO 2 2 m 2 sec 1 ), T s is soil temper ature (C), M s is soil volumetric moisture content (m 3 /m 3 ), Mean Temp is the mean monthly air temperature (C), Models other than M s that had fit less than R 2 = 0.10 were not reported.

PAGE 59

59 Table 2 9. Results of nonlinear models of soil CO 2 efflux rates u sing soil temperature as a predictor FRI Model Q 10 R 2 p 1YR R s = 0.9727 e 0.0502*Ts 1.65 0.60 < 0.001 2YR R s = 0.7973 e 0.0673*Ts 1.96 0.80 < 0.001 40YR R s = 0.9712 e 0.0770*Ts 2.16 0.76 < 0.001 Models are of monthly mean soil CO 2 efflux rate ( R s 2 m 2 sec 1 ) responses to soil temperature (T s ). Data are presented by prescribed fire return interval (FRI). Coefficients were estimated using statistical software JMP 9.0. Q 10 was calculated using the exponential equation Q 10 = e (Lundeg initial model. R 2 is the linear regression fit of the exponentially predicted R s values against the observed R s values and p is the significance of the model fit of predicted to observed values.

PAGE 60

60 Table 2 10. Linear regression relationship between soil CO 2 efflux rates and soil temperature by fire return interval and season FRI Season Model and estimates R 2 p 1YR Spring R s = 1.0261 + 0.0700* T s 0.29 0.021 Summer R s = 12.83.11 0.2799* T s 0.40 0.068 Fall R s = 1.4494 + 0.2159* T s 0.71 < 0.001 Winter R s = 0.4895 + 0.14355* T s 0.69 0.002 2YR Spring R s = 1.0808 + 0.09766* T s 0.48 0.002 Summer R s = 15.7301 0.3751* T s 0.28 0.144 Fall R s = 3.1609 + 0.3197* T s 0.90 < 0.001 Winte r R s = 0.4894 + 0.1661* T s 0.71 0.002 40YR Spring R s = 1.5400 + 0.1302* T s 0.11 0.170 Summer R s = 49.5720 1.6300* T s 0.63 0.011 Fall R s = 4.3665 + 0.4270* T s 0.71 < 0.001 Winter R s = 2.0244 + 0.3544* T s 0.90 < 0.001 Spring (March May), s ummer (June August), fall (September November), and winter (December February). R s is soil CO 2 2 m 2 sec 1 ), T s is soil temperature (C).

PAGE 61

61 Table 2 11. S easonal nonlinear models of soil CO 2 efflux rates using soil temperature as a predictor FRI Season Equation Q 10 R 2 p 1YR Spring R s = 1.5095 e 0.0238*Ts 1.27 0.26 0.031 Summer R s = 37.6590 e 0.0721*Ts 0.49 0.45 0.049 Fall R s = 0.8564 e 0.0598*Ts 1.82 0.66 < 0.001 Winter R s = 0.2704 e 0.1200*Ts 3.32 0.77 0.000 2YR Spring R s = 0.2778 e 0.1250*Ts 3.49 0.33 0.012 Summer R s = 36.2629 e 0.0692*Ts 0 .50 0.27 0.149 Fall R s = 0.5967 e 0.0823*Ts 2.28 0.89 < 0.001 Winter R s = 0.3701 e 0.1114*Ts 3.05 0.78 0.001 40YR Spring R s = 2.2967 e 0.0292*Ts 1.34 0.10 0.195 Summer R s = 259.9167 e 0.1388*Ts 0.25 0.62 0.012 Fall R s = 0.6935 e 0.08 83*Ts 2.42 0.69 < 0.001 Winter R s = 0.3990 e 0.1383*Ts 3.99 0.96 < 0.001 Models are of monthly mean soil CO 2 efflux rate ( R s 2 m 2 sec 1 ) responses to soil temperature (T s ). Data are presented by prescribed fire return interval (FRI). Coeffic ients were estimated using statistical software JMP 9.0. Q 10 was calculated using the exponential equation Q 10 = e coefficient estimated in the initial model.

PAGE 62

62 Table 2 12. S easonal linear models of so il CO 2 efflux rates using soil moisture content as a predictor FRI Season Equation R 2 p 1YR Spring R s = 2.7146 1.0286*M s 0.01 0.675 Summer R s = 3.8796 + 4.5559*M s 0.13 0.475 Fall R s = 2.2962 + 7.8960*M s 0.15 0.109 Winter R s = 1.0143 + 0.5351*M s 0. 00 0.904 2YR Spring R s = 3.4316 2.5191*M s 0.08 0.264 Summer R s = 4.2300 + 7.2134*M s 0.44 0.152 Fall R s = 2.5987 + 10.2822*M s 0.14 0.129 Winter R s = 2.4157 3.7664*M s 0.05 0.366 40YR Spring R s = 3.9373 + 0.6649*M s 0.00 0.887 Summer R s = 4.3603 + 28.4285*M s 0.82 0.013 Fall R s = 2.1477 + 28.9752*M s 0.49 0.001 Winter R s = 2.8535 2.0522*M s 0.02 0.627 Spring (March May), summer (June August). fall (September November), winter (December February). R s is soil CO 2 2 m 2 sec 1 ), M s is soil moisture content (m 3 /m 3 ).

PAGE 63

63 Table 2 13. Step wise multiple regression models to explain soil CO 2 efflux rates using field parameters FRI Model R 2 RMSE p All FRI R s = 2.982 + 0.243*M Temp + 8.416*M s + 1.178*Duff Depth 0.80 0.78 < 0.001 1YR R s = 0.766 + 0.209*M Temp + 5.103* M s 0.334*Dist Nearest 0.86 0.52 < 0.001 2YR R s = 3.312 + 0.255* T s + 6.990* M s + 0.049*P BA 0.89 0.50 < 0.001 40YR R s = 2.292 + 0.421* T s + 16.805* M s 2.512*Dist Nearest 0.82 0.86 < 0.001 R s is soil CO 2 eff 2 m 2 sec 1), FRI is prescribed fire return interval treatment type. For model term details see Table 2 1. Terms were selected for inclusion using a forward step wise procedure in SAS JMP 9.0 (SAS Institute Inc., Cary, North Carolina, US A) based on input parameter significance (p < 0.05) and minimum model BIC.

PAGE 64

64 Figure 2 1. The research site, Tall Timbers Research Station was located in Leon County, Florida, USA. The site is approximately 30 km north of the city of Tallahassee, Flor ida, USA. Map produced by David Godwin.

PAGE 65

65 Figure 2 2. Ground (left) and aerial (right) images of three of the soil CO 2 efflux sampling plots located within the Tall Timbers Research Station in Leon County, Florida, USA. The top images show an annual b urn frequency site (1YR), the middle images a two year burn frequency site (2YR), and the bottom image a site unburned since 1966. Ground images original to the author. Ground photo graphs courtesy of David Godwin Aerial images courtesy of Mic rosoft Bing Maps.

PAGE 66

66 Figure 2 3. PVC soil CO 2 efflux rate ( R s ) sampling collar (20 cm diameter) installed at a frequently burned plot at the Tall Timbers Research Station near Tallahassee, Florida, USA. Photo graph courtesy of David Godwin

PAGE 67

67 Figure 2 4. Monthly mean soil temperature ( C) and monthly mean regional 2 m air temperature (Mean Air Temp) for three prescribed fire treatment types at the Tall Timbers Research Station near Tallahassee, Florida, USA. Soil temperature data were from 27 sampl e points per treatment type, measured three times daily once per month. Mechanical difficulty resulted in erroneous data collected in January 2009 and February 2011. Data were not recorded in the month of October 2010. Monthly mean air temperature were recorded hourly at the Florida Automated Weather Network (FAWN) station approximately 30 km away in Quincy, Florida.

PAGE 68

68 Figure 2 5. Mean soil CO 2 efflux rate ( mol CO 2 m 2 sec 1 ), soil temperature (C), and soil moisture content (m 3 /m 3 ) per prescribed f ire treatment type at Tall Timbers Research Station near Tallahassee, Florida, USA. Data represent overall study area means of monthly measurements from August 2009 until May 2011. Letters indicate significant differences between treatment types assessed using repeated measures ANOVA and Tuke

PAGE 69

69 Figure 2 6. Monthly mean soil moisture content (m 3 /m 3 ) for three prescribed fire treatment types at the Tall Timbers Research Station near Tallahassee, Florida, USA. Mechanical difficulty resulted in erroneous data collected in August 2009. Data were not recorded in the month of October 2010.

PAGE 70

70 Figure 2 7. Monthly regional Palmer Drought Severity Index (PDSI) scores from the National Oceanic and Atmospheric Administration (NOAA) National Climatic Data Center (NCDC). All scores below zero represent drought conditions for the northwest Florida region.

PAGE 71

71 Figure 2 8. Monthly mean soil CO 2 efflux rates ( mol CO 2 m 2 sec 1 ) for three prescribed fire treatment types at the Tall Timbers Research Station near Tallahassee, Florida, USA. Data were from 27 sample points per treatment type, measured three times daily once per month. Data were not recorded in th e month of October 2010.

PAGE 72

72 Figure 2 9. Linear regression of the relationships between mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and mean: stand density (trees per ha 1 ), distance to nearest tree (m), stand basal area (m 2 ha 1 ), and diameter at breast height of the nearest tree (cm). Each point represents entire study period means per sample plot with all treatments combined.

PAGE 73

73 Figure 2 10. Linear regression of the relationships between mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and mean: duff depth (cm), litter depth, and total litter+duff depth (cm). Each point represents entire study period means per sample plot with all treatments combined.

PAGE 74

74 Figure 2 11. Linear regression of the relationships between monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and monthly mean soil temperature (T s ) ( C) for three prescribed fire intervals at the Tall Timbers Research Station near Tallahassee, Florida, USA. Each point represents monthly mean values per sample plot.

PAGE 75

75 Figure 2 12. The r elationship between monthly mean soil CO 2 2 m 2 sec 1 ) ( R s ) and monthly mean soil temperature (C) (T s ) as modeled using an exponential equation (Equation 2 2). Data presented are from sites at the Tall Timbers Research Station near T allahassee, Florida, USA, representing three prescribed fire treatment intervals. Each point represents monthly mean values per sample plot.

PAGE 76

76 Figure 2 13. Linear regression of the relationships between monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and monthly mean air temperature (M Temp) ( C) for three prescribed fire intervals at the Tall Timbers Research Station near Tallahassee, Florida, USA. Monthly air temperature data were from the Florida Automated Weather Network Station (FAWN S) at Quincy, Florida, USA. Each point represents monthly mean values per sample plot.

PAGE 77

77 Figure 2 14. The relationship between monthly mean soil CO 2 2 m 2 sec 1 ) ( R s ) and monthly mean air temperature (C) (M Temp) as modeled using an exponential equation (Equation 2 2). Data presented are from sites at the Tall Timbers Research Station near Tallahassee, Florida, USA, representing three prescribed fire treatment intervals. Monthly air temperature data were from the Florida Automated Weather Network Station (FAWNS) at Quincy, Florida, USA. Each point represents monthly mean values per sample plot.

PAGE 78

78 Figure 2 15. Linear regress ion of the relationships between monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and monthly mean soil moisture content (M s ) (m 3 /m 3 ) for three prescribed fire intervals at the Tall Timbers Research Station near Tallahassee, Florida, USA. Each point represents monthly mean values per sample plot.

PAGE 79

79 Figure 2 16 Predicted monthly total soil carbon flux from August 2009 to July 2010 by prescribed fire return interval (FRI). Flux values were predicted using FRI specific linear models of soil CO 2 efflux rate responses to 2 m ambient air temperature.

PAGE 80

80 CHAPTER 3 T HE EFFECT S OF LITTER INPUTS AND PRESCRIBED FIRE ON S OIL CO 2 EFFLUX RATES IN NORTH FLORI DA OLD FIELD FORESTS Background Forests and forest soils are a significant repository of global carbon and represent ol (Post et al., 1982; Luo and Zhou, 2006) In recent years forest management practices that reduce carbon emissions and increase carbon sequestration have been identified as a potential pathway towards reducing global atmospheric CO 2 concentrations (McKi nley et al., 2011 ; Maier et al., 2012 ) One way that management practices have shown to influence forest carbon emissions is through their e ffect on soil CO 2 efflux rates (Luo and Zhou, 2006) Soil CO 2 efflux (R s ) represents one of the dominant fluxes of CO 2 from forested systems to the atmosphere, with the estimated global annual R s flux an order of magnitude greater than total anthropogenic emissions from fossil fuel combustion (Schlesinger and Andrews, 2000; Luo and Zhou, 2006) Given the magnitude of the R s flux and the significant role that forest soils play in global carbon dynamics it is important to understand the connections between forest management practices and the factors that drive soil CO 2 efflux rates. Soil CO 2 efflux (R s ) is a combination of CO 2 respired by plant roots a nd associated rhizosphere fungi (R a ) and heterotrophic soil microorganisms (R h ). Methods of partitioning the relative contribution of R h and R a sources to R s have been discussed in detail recently by several authors (Hanso n et al., 2000; Kuzyakov, 2006; Subke et al., 2006) It has been recognized that one of the greatest challenges to partitioning sources of R s is the delineation among sources in the field without disturbing the soil matrix and biota (Fenn et al., 2010) In place of invasive field methods of partitioning such as trenching and girdling, or ex situ methods such as laboratory incubation, some

PAGE 81

81 studies have investigated the specific response of R h relative to R s by manipulating labile carbon inputs (Cleveland e t al., 2006; Salamanca et al., 2006; Zimmermann et al., 2009; Chemidlin Prvost Bour et al., 2010) While litter manipulation treatments are not substitutions for experimental partitioning methods, they can reveal information about the importance of abov eground inputs on soil CO 2 efflux without heavily impacting soil biotic and abiotic conditions. It has been shown that litterfall and leaf litter can be a source of carbon for the heterotrophic soil microbes responsible for R h fluxes ( Nahdelhoffer et al., 2004; Luo and Zhou, 2006 ; Sayer, 2006 ) Understanding the R s and R h response to variable vegetative aboveground inputs and management regimes is relevant for understanding and predicting soil carbon dynamics under changing climate and vegetation assemblag es (Sayer, 2006) Improving the understanding of these mechanisms is particularly important as studies have suggested that increased atmospheric CO 2 and anthropogenic N deposition may lead to increased litterfall and altered soil carbon fluxes as heterotr ophic soil microorganisms respond to changing carbon inputs (Zak et al., 2003; Quinn Thomas et al., 2009) Many previous studies following direct leaf and needle litter additions have reported increased R s rates relative to control s with the response s att ributed to increased R h contributions to R s (Bowden et al., 1993; Jonasson et al., 2004; Salamanca et al., 2006; Sulzman et al., 2012) Aerobic soil microbial populations have also been shown to increase metabolic activity when presented with more targete d labile carbon additions such as: glucose, amino acids, root exudates, and dissolved organic matter (DOM) (Nobili et al., 2001; Cleveland et al., 2006) Temporal responses

PAGE 82

82 of soil microbial populations to carbon additions have proven to be relatively sho rt, with elevated soil CO 2 efflux rates observed within ~12 hours of treatment with elevated rates lasting many days or months (Nobili et al., 2001; Cleveland et al., 2006; Chemidlin Prvost Bour et al., 2010) Seasonal variation s in litter inputs to the soil have also been shown to influence R s rates, with increased rates observed in the autumn season despite decreased aboveground photosynthetic activity and soil temperature (Kutsch et al., 2010) Such autumnal increases in R s have been attributed to the availability of labile carbon from deciduous leaf senescence during the autumn months (Kutsch et al., 2010) Previous studies have found litter exclusion to reduce R s rates due to a reduction in available carbon for soil microbial metabolism. Li et al (2004) reported that prolonged (7 years) litter exclusion in a natural Pinus stand in Puerto Rico reduced (54%) in situ R s rates as well as soil microbial biomass (67%) relative to control. Similarly, in a hardwood forest in North Carolina, USA, Reynold s and Hunter (2001) found that litter exclusion during a six month study significantly reduced R s rates relative to control. In a deciduous hardwood forest in Massachusetts, USA, Bowden et al. (1993) determined following a six month litter exclusion exper iment that the decomposition of recent leaf litter represented approximately 12% of total R s rates. Our study sought to address the response of R s as a proxy for R h to changes in aboveground litterfall within forest stands representing three prescribed fir e management regimes: annual prescribed fire (1YR), biennial prescribed fire (2YR) and fire exclusion (40YR). The intention of this study was to assess the relative influence of inputs of leaf litter as a labile carbon source for heterotrophic soil microo rganism

PAGE 83

83 metabolism within each common fire management type. The manipulation of labile carbon inputs was intended to give insight into the magnitude, temporal, and seasonal response of the heterotrophic soil microorganisms to leaf litter carbon inputs. W hile not directly partitioning sources of R s this experiment specifically targeted the R h component of R s without invasive or destructive procedures. This study seeks to address the hypothesis that aboveground litter influences heterotrophic soil microbi al decomposition and soil CO 2 efflux rates in both frequently burned and long unburned old field forests. Methods Study S ite The study sites were located within the Tall Timbers Fire Ecology Research Plots ( Stoddard Fire Research Plots ) at the Tall Tim bers Research Station (TTRS) in Leon County, Florida, USA, approximately 30 km from the cities of Tallahassee, Florida (to (Figure 3 1) ( Clewell and Komarek, 1975; Glitzenstein et al ., 2012) The Stoddard Fire Research Plots were established in the 1960 s as a long term study of the influence of prescribed fire frequency on old field forest vegetation and soils (Clewell and Komarek, 1975; Glitzenstein et al., 2012) Prior to establis hment of the plots most of the region was burned annually to improve hunting since the late 1800s and 1920s (K Robertson, 2012 pers. comm.) For this study, sampling took place within the annually burned (1YR), biennially burned (2YR) and fire excluded (4 0YR) Stoddard Fire Research Plots (Figure 3 2). The study site elevation was approximately 59 m a.s.l. Average annual precipitation was 137 cm with the majority falling during the summer months of June, July and August (National Climate Data Center 2009, Thomasville, Georgia). Mean

PAGE 84

84 maximum and minimum temperatures for January and July for the area from long term records (1971 2000) were 16.8C and 4.6C for January and 33C and 21.8C for July (National Climate Data Center 2009, Thomasville, Georgia). S oils within the sites were heavily cultivated for corn and cotton from the 1820s 1920s and occasionally as recently as the 1950s, with subsequent soil and vegetation assemblages highly influenced by past agricultural practices (Clewell and Komarek, 1975) Soils were generally classified as fine loamy, kaolinitic, thermic Typic Kandiudults of the Orangeburg and Faceville series (Natural Resource Conservation Service (NRC) Soil Survey Geographic Database (SSURGO) ) Vegetation across the frequently burned si tes consisted of an overstory mixture of naturally regenerated shortleaf pine ( Pinus echinata P. Mill), and loblolly pine ( P. taeda L.) and an understory composed of annual grasses and hardwood resprouts (Clewell and Komarek, 1975; Myers and Ewel, 1990; En gstrom and Palmer, 2005; Glitzenstein et al., 2012) Vegetation within the unburned sites consisted of a mixture of shortleaf pine, loblolly pine, sweetgum ( Liquidambar styraciflua L.), mockernut hickory ( Carya alba (L.) Nutt. ex Ell.), live oak ( Quercus virginiana P. Mill.) and water oak ( Q. nigra L.) (Clewell and Komarek, 1975; Myers and Ewel, 1990) Litter M anipulation a nd S ampling The seven month litter manipulation and R s sampling experiment was established in May 2011. From Ju ne 2011 until December 2011, soil CO 2 efflux rate s (R s CO 2 m 2 sec 1 ), soil temperature (T s ) ( C), and soil volumetric moisture content (M s ) (m 3 /m 3 ) were sampled three times a day once per month. Sampling took place within a total of nine plots e stablished within three blocks, with each block consisting of a representative plot of three prescribed fire return interval (FRI) treatment types: annual

PAGE 85

85 burn (1YR), biennial burn (2YR) and long unburned (40YR) (Figure 3 2 ). To account for variability wi thin the plot, each individual plot was comprised of nine 20 cm diameter PVC sample soil collars arranged in a 3 x 3 grid with 5 m separation following Kobziar and Stephens (2006). PVC sampling collars were constructed of Schedule 30 white 20 cm diameter pipe cut to 10 cm lengths and beveled along one edge. Collars were inserted beveled edge down into the soil or duff to a depth of approximately 8 cm using a rubber mallet. All collars were installed at least four weeks prior to the start of sampling to a llow any soil disturbance from installation to normalize. During the course of study, any vegetative growth within the sample collars was clipped and removed prior to R s measurement. Two experimental litter manipulations plus control were randomly assigne d to collars within each plot: litter addition, (3 collars), litter exclusion (3 collars), and control (3 collars). Around each sam pling collar a low rectangular frame measuring approximately 40 cm in length on each side and rising approximately 5 cm abov e the soil surface was installed (Figure 3 3 x 5.08 cm pine lumber and was the unit of litter addition, exclusion, or control. The interior of each treatment box measured 0.16 m 2 The treatment box was d esigned to expand the area of litter manipulation beyond the confine of the PVC soil collar to include the area outside the collar where T s and M s measurements were taken. This was important as it was anticipated that the litter manipulations might influe nce microsite T s and M s conditions that would otherwise be unaccounted for if the unit of treatment were restricted to within the soil collar. The litter addition sample frames received a one time addition of 140 g. of litter (the equivalent to 8750 kg ha 1 ) during the

PAGE 86

86 spring season (May) (Figure 3 3). This litter addition represented a 47%, 38%, and 37% increase over the mean annual litterfall load at the 1YR, 2YR, and 40YR sites respectively (K. Robertson, unpublished data; Reid et al., 2012). To ensur e that the litter addition composition was representative of natural inputs, fresh leaf litter material was gathered in each plot during the preceding autumn (2010) The gathered material represented a natural mixture of foliar material mostly from Pinus Cary a and Quercus species. Leaf litter was bagged and oven dried at 50C for one week to achieve uniform moisture content. It was then ground using a 5 mm sieve on a Wiley Mill (Thomas Scientific Inc., Swedesboro, NJ, USA) to improve microbial availabi lity. The freshly ground litter was then applied one time within three randomly selected sample frames per plot. Care was taken to evenly distribute the litter within the entire area of the sample frame including within the PVC collar. Fresh litter in put was excluded from three randomly selected treatment boxes per plot via the construction of an enclosure of flexible plastic mesh stapled to wooden survey stakes (Figure 3 3 ). This design facilitated the exclusion of vertical and horizontal litterfall, while still allowing for the free flow of air, precipitation, and sunlight; as well as access for the R s T s and M s measurements taken with the sampling instrument and sensor probes. Litterfall material collected on top of the exclusions was removed and discarded monthly to prevent shading of the treatment box. Three randomly selected treatment boxes per plot did not receive any manipulation and remained throughout the study as the experimental control. Sampling of R s for all plots was conducted using a LI COR Biosciences LI 8100 automated soil CO 2 sampling instrument with a 20 cm survey chamber (LI COR

PAGE 87

87 Biosciences Inc., Lincoln, NE, USA) (Figure 3 4 ). Concurrently with R s measurements, soil temperature (T s ) and moisture content (M s ) at 10 cm and 5 cm d epth s, respectively were recorded onboard the LI 8100 using an Omega 8831 type E T Handle temperature probe and a Decagon Systems EC 5 soil moisture probe (Omega Inc., Stamford, CT; Decagon Systems Inc., Pullman, WA). On the monthly sampling day, each tr eatment box was sampled three times: once in the morning, once at mid day, and once again in the evening hours. A total of 243 measurements were taken per month (nine collars x three daily measurements x three fire treatment types x three replicates). Th e resulting dataset for the entire seven month study totaled 1655 R s measurements after exclusions due to hazardous weather and equipment malfunctions. Problems with sampling equipment resulted in erroneous soil moisture content measurements during the mo nth of June and erroneous soil temperature measurements during the month of August. Recorded soil moisture content values less than 0.00, and soil temperature measurements greater than 40 C were ignored in analyses. Plot level forest vegetative and fores t floor characteristics were assessed in the winter and spring of 2011. Vegetation was sampled using a 15 m radius circular plot (0.07 ha) centered on the middle R s sample collar. The following field parameters with abbreviation and unit were recorded on e time per plot: basal area (BA) (m 2 ha 1 ) hardwood basal area ( HW BA) (m 2 ha 1 ) pine basal area ( P BA) (m 2 ha 1 ) and stand density (TPH) (trees ha 1 ) Plot level mean litter depth (Litter) (cm) and mean duff depth (Duff) (cm) came from the averages of measurements taken as part of a previous study in the Stoddard Fire Plots reported in Chapter 2 (this document).

PAGE 88

88 Due to the technical difficulties that led to gaps in measurements taken by the on site weather station, monthly mean ambient temperature (M T emp) measurements and precipitation totals were recorded by the Florida Automated Weather Network (FAWN) site at Quincy, Florida, approximately 30 km from Tall Timbers Research Station During the period of study from June through December 2011, mean mont hly ambient temperatures (M Temp) ranged from 13.67 27.82 C (Figure 3 5 ) and total monthly precipitation ranged from 4.50 20.55 cm. In plot measured soil temperature ranged from 8.91 34.75 C, while in plot measured soil moisture content ranged fr om 0.00 0.32 m 3 /m 3 Regional precipitation totals during and prior to the start of the study were Drought Severity Index (PDSI)(Figure 3 6 )(National Oceanic and At mospheric Administration, National Climatic Data Center). Analysis To determine the overall effect of treatments, a repeated measures analysis of variance (ANOVA) was used to examine the effect s of prescribed fire inter val (FRI), litter treatment type tim e (sample month), and interaction effects on plot level monthly mean soil CO 2 efflux rates, soil temperature, and soil moisture content. Significant treatment effects were identified at p value < 0.05. Where significant effects were identified, differenc s To determine the effects of litter treatments on R s T s and M s within specific FRIs, additional repeated measures analysis of variance (ANOVA) tests were performed for each FRI, with signif icant differences among litter treatment types analyzed using To assess for differences between plot level forest and stand

PAGE 89

89 characteristics by FRI, one way ANOVA tests were used. Where significant differences were identified, differenc To examine the relationships between R s rates and T s M s and monthly mean air temperature (M Temp) and precipitation, linear (Equation 3 1) regression models were developed using monthly plot mea ns per FRI and litter treatment. In addition, non linear models of the relationship s between monthly mean R s rates and T s and M Temp were explored using an exponential equation (Equation 3 2) frequently used to describe the response of R s rates to soil te mperature (Lundegardh, 1927; Samuelson et al., 2004; Concilio et al., 2005; Kobziar and Stephens, 2006). ( 3 1) or ( 3 2) ( 3 3 ) 0 and 1 we re coefficients estimated through regression analysis. Residuals of regressions were checked for normality and heteroscedasticity. In addition to those mentioned previously, an exponential equation (Equation 3 3) describing the response of soil CO 2 efflux to a 10 C change in soil temperature was developed per FRI and litter treatment type (Lundegardh, 1927; Samuelson et al., 2004; Concilio et al., 2005; Kobziar and Stephens, 2006). All statistical analyses were performed u sing JMP 9.0 (SAS Institute, Cary, NC, USA). Results Plot level forest conditions and composition varied significantly between prescribed fire treatment types (FRI). Stand density ( 1716.41 trees ha 1 ) basal area (37.72 m 2 ha 1 ) duff (1.58 cm) and litter depth (2.81 cm ) were all greatest in the 40YR

PAGE 90

90 FRI and lowest in the 1YR FRI (282.93 trees ha 1 11.79 m 2 ha 1 0.08 cm, 1.77 cm, respectively) (Table 3 2). Both the pine and hardwood species components of total basal area increased from the 1YR to the 40 YR FRI, with the most hardwood (15.73 m 2 ha 1 ) and pine (21.99 m 2 ha 1 ) basal area in the 40YR FRI and the least hardwood and pine basal area in the 1YR FRI (3.87 m 2 ha 1 and 7 .92 m 2 ha 1 respectively) (Table 3 2). During the seven month study period mon thly mean R s rates varied considerably: 1YR: control (0.95 6.15 mol CO 2 m 2 sec 1 ) litter addition (1.24 8.15 mol CO 2 m 2 sec 1 ), and litter exclusion (0.92 6.44 mol CO 2 m 2 sec 1 ) ; 2YR: control (0.88 7.36 mol CO 2 m 2 sec 1 ) litter addition (1.45 10.42 mol CO 2 m 2 sec 1 ), and litter exclusion (1.17 7.15 mol CO 2 m 2 sec 1 ) ; 40YR: control (1.85 11.30 mol CO 2 m 2 sec 1 ) litter addition (2.17 12.63 mol CO 2 m 2 sec 1 ), and litter exclusion (1.60 10.53 mol CO 2 m 2 sec 1 ) In genera l, in all FRIs and litter treatments, soil moisture content (M s ) (m 3 /m 3 ), soil temperature (T s ) ( C) and R s rates varied by sample month, with the highest T s and R s observed in the summer months and the lowest in the fall and winter months (Table 3 3) (Fig ure s 3 7 and 3 8). R s and T s closely followed seasonal trends in ambient air temperature (Figure 3 5), with monthly mean T s in each litter treatment and FRI highly positively linearly correlated with monthly mean ambient air temperature (R 2 = 0.92 0.95) (Table 3 10). Soil moisture content did not follow the same seasonal trends as R s and T s with the highest M s observed in all FRIs and litter treatments in the summer and winter months and the lowest M s observed during the months of September and October (Figure s 3 7 and 3 8). Soil moisture content throughout the study period was not significantly correlated with monthly precipitation (Figure 3 5) (Table 3 13) and was

PAGE 91

91 likely influenced by the effects of an on going regional drought that started prior to and continued throughout the entire study period (Figure 3 6). Analysis of T reatment E ffects Repeated measures ANOVA was used to identify significant overall treatment effects (p < 0.05) due to FRI, litter treatment type, sample month (tim e), and the interaction of FRI x month, litter x month, and FRI x litter (Table 3 3). When litter treatments were ignored and the data pooled, significant differences in monthly mean R s rates (F = 24.20 p < 0.0001), soil temperature (T s ) (F = 6.80 p = 0.0 016), and soil moisture content (M s ) (F = 43.06 p < 0.0001) were found between the prescribed fire (p < 0.05) found that monthly mean R s rates were significantly highe r in the 40YR FRI than in the 2YR and 1YR FRI (Table s 3 3 and 3 4). Similar tests found that monthly mean M s was significantly lower in the 40YR FRI, while monthly mean T s was significantly higher in the 1YR than the 40YR FRI (Table 3 4). The treatment e ffect of FRI on R s (F = 3.92 p < 0.001), T s (F = 14.73 p < 0.001), and M s (F = 2.18 p = 0.0245) varied significantly over time (Table 3 3) (Figure 3 8), with similar monthly and seasonal variations in R s T s and M s observed and discussed in Chapter 2 (thi s document). When FRI was ignored and the effect of litter treatment type assessed exclusively, significant treatment effects on monthly mean R s (F = 11.29 p < 0.0001) and M s (F = 7.00 p = 0.0014) were identified (Table 3 overall, litter addition treatments resulted in significantly higher monthly mean R s rates than the exclusion and control treatments, while the litter exclusion treatments resulted in significantly lower overall monthly mean M s relative to the litter addi tion and control treatments (Table 3 5). Monthly mean s oil temperature was not significantly influenced

PAGE 92

92 by litter manipulation trea tments when FRI was ignored The treatment effect of litter manipulation only varied significantly with time (F = 1.51 p = 0.0402) for M s (Table 3 5) (Figure 3 7). While the overall trends in FRI and litter manipulation described previously (Table 3 3) provide insight into the broader effects of such treatments, the primary interest of this study was to identify the effects o f litter manipulation treatments (litter addition, litter exclusion and control ) on R s T s and M s within each FRI (1YR, 2YR, and 40YR) (Table s 3 6 and 3 7). To accomplish this, separate repeated measures ANOVA and and variable of interest (R s T s and M s ) (Table 3 6). Within the 1YR FRI, the control 2 m 2 sec 1 ) and litter exclusion treatment 2 m 2 sec 1 ) mean R s rates were significantly lower (28%) than the litter addition treatment (4.2 2 m 2 sec 1 ) Within the 2YR FRI, the control 2 m 2 sec 1 ) treatment mean R s rates were significantly lower (32%) than the litter addition treatment 2 m 2 sec 1 ) (Table 3 6) (Figures 3 9 and 3 10) Litter manipulation h ad no significant effect on R s rates in the 40YR FRI. Soil moisture content (M s ) varied by litter treatment only within the 1YR prescribed fire treatment (Table s 3 6 and 3 7) (Figure s 3 11 and 3 12) In the 1YR FRI litter exclusion (0.12 m 3 /m 3 ) reduced soil moisture content relative to the control (0.16 m 3 /m 3 ) but not the litter addition (0.14 m 3 /m 3 ) (Table 3 6). No significant differences in soil temperature (T s ) were found among litter treatment types within the FRI sites (Table s 3 6 and 3 7 ) (Figure s 3 13 and 3 14) Effects of Treatments on the Response of R s to Abiotic Factors Simple linear regression models (Equation 3 1) and non linear (Equation 3 2) exponential models were developed by litter treatment type and f ire return interval to

PAGE 93

93 assess the influence of litter treatments on the relationships between R s rates and soil temperature (Table s 3 8 and 3 9). The l inear regression models indicated that positive relationship s between monthly mean R s rates and T s exist ed for all fire return intervals and litter treatment types (R 2 = 0.38 0.61) (Table 3 8) ( Figure s 3 1 5, 3 16, and 3 17 ) Litter manipulation appeared to slightly weaken the relationships between R s and T s in all FRIs. In general, the regression coeffic ients of all R s and T s models were slightly lower than those developed in Chapter 2 of this document. The non linear exponential models also indicated that positive relationships between monthly mean R s rates and T s existed for all fire return intervals a nd litter treatment types (R 2 = 0.36 0.58) (Table 3 9) (Figure 3 18, Figure 3 19, and Figure 3 20). Like the linear regression models, litter manipulation generally appeared to slightly weaken the relationships between R s rates and T s with the highest r egression coefficients in the control litter treatments in the 1YR and 2YR FRIs and the litter exclusion treatment in the 40YR FRI (Table 3 9). Using the 1 estimate from Equation 3 2 in the Q 10 model (Equation 3 3), the response of R s rates to 10 C changes in T s (Q 10 value) were calculated for each litter treatment type and FRI (Q 10 = 1.5 7 3. 40 ) (Table 3 9). In the 2YR and 40YR FRIs, Q 10 values were hi ghest in the litter addition treatments ( Q 10 = 2 10 and Q 10 = 3 40, respectively) while in the 1YR FRI, Q 10 values were highest in the control treatment (Table 3 9). Additional simple linear regression models using Equation 3 1 were developed per litter tr eatment type and FRI to assess the relationships between monthly mean soil moisture content (M s ), monthly total precipitation (Precip), and monthly mean R s rates. In all litter treatment and FRI models, no significant relationship s were found between R s r ates and M s (R 2 = 0.00 0.18 p > 0.05) (Table 3 11). These results are similar to

PAGE 94

94 those of Chapter 2 (this document) wherein M s explained little of the temporal variation in monthly mean R s rates (R 2 = 0.03 0.14). In contrast, simple linear regression models (Equation 3 1) by litter treatment type and FRI of the relationships between monthly mean R s rates and monthly total precipitation (Precip) identified strong linear relationships in all models (R 2 = 0.46 0.75) (Table 3 12). Further analysis foun d that the monthly pattern of precipitation during the study period followed similar seasonal variations in soil and ambient air temperature (Figure 3 5) that resulted in T s and Precip being strongly correlated (R 2 = 0.58 0.64) (Table 3 14). This multic ollinearity between T s and Precip limited further interpretation of the effects of monthly precipitation patterns on the temporal variations in monthly mean R s rates. Discussion Many of the studies of soil CO 2 efflux rates available for comparison in the s outheastern US took place in industrial plantation forests and as such some differences in results were not surprising (Ewel et al., 1987a; Ewel et al., 1987b; Fang et al., 1998; Gough and Seiler, 2004; Samuelson et al., 2004; Gough et al., 2005). The ran ge of monthly mean R s rates observed in the litter control units in this study were similar but generally higher than those reported elsewhere in several other studies of southeastern US soil CO 2 efflux rates (Maier and Kress, 2000; Gough and Seiler, 2004; Maier et al., 2004; Gough et al., 2005; Samuelson et al., 2004; Samuelson and Whitaker, 2012). For example, in a study of a Georgia, USA, loblolly pine plantation Samuelson et al. (2004) reported R s rates ranging from 1 6 mol CO 2 m 2 sec 1 while Butno r et al. (2003) reported R s rates ranging from 2.23 2 m 2 sec 1 in a loblolly pine plantation in North Carolina, USA. In another example, monthly mean R s rates from a yearlong study in a similarly structured frequently burned, natural longle af

PAGE 95

95 pine forest in Coastal Alabama, USA ranged from 1.6 2 m 2 sec 1 (Samuelson and Whitaker, 2012). Not surprisingly the range of R s rates in the litter control treatments in this study were similar to those reported in Chapter 2 of this docum ent following a nearly two year survey of monthly and seasonal variability of R s rates in the Stoddard Fire Plots. It is possible that the higher R s rates observed in this study in comparison to those cited previously may have been due to differences in t he soil CO 2 efflux measurement systems. While research by Madsen et al. (2008) has shown that there are no significant differences in estimated R s flux rates between the commonly used LI 6400 and LI 8100 instruments, previous research by Heinemeyer and Mc Namara (2011) has shown that soil CO 2 efflux measurement chamber type can significantly influence measured flux rates, with closed static chambers consistently underestimating R s fluxes (21 39%) compared to closed dynamic chambers like the LI 8100 instrume nt used in this study (LI COR Biosciences, Inc. Lincoln, NE, USA). Effect of P rescribed F ire M anagement Our results found that the three prescribed fire management methods assessed in this study (1YR, 2YR, and 40YR) significantly altered R s rates regardless of litter manipulation type, with the highest mean R s rates in the long unburned (40YR) sites and the lowest in the 1YR sites. While temporal variations in R s rates across all treatments were generally well explained by positiv e correlations with soil temperature (T s ), neither T s nor M s explained the differences in R s rates between FRIs. These results were consistent with those reported in Chapter 2 (this document). The observed differences in monthly mean R s rates between FRI sites were likely due to variations in stand level biomass and productivity (Raich and Tufekciogul, 2000), vegetative composition (Wang et al., 2006), and disturbance history (Hanula et al., 2012) that

PAGE 96

96 facilitated higher R a and R h contributions to total R s in the 40YR sites than the 1YR and 2YR sites. See Chapter 2 of this document for further discussion of the variations in mean monthly R s rates between the 1YR, 2YR, and 40YR Stoddard Fire Plots. Effect of L itter A ddition The results of ou r litter manipulation found that R s rates in the frequently burned sites responded positively to the litter supplements as litter additions led to significant increases (28% and 32%) in R s 2 m 2 sec 1 ) and 2YR 2 m 2 sec 1 2 m 2 sec 1 2 m 2 sec 1 respectively). Other studies of litter manipulation have reported results similar to those described here. For example, in a litter addition e xperiment in the Cascade Mountains of Oregon, USA, Sulzman et al. (2012) reported a 34% increase in R s rates relative to control following conifer needle litter additions. In another study, Chemidlin Prvost Bour et al. (2010) reported mean R s rates incr eased 60 120% relative to controls following litter additions in a field study in France It is important to consider that the elevated R s rates observed following litter additions in the 1YR and 2YR treatments in our study may have been artificially influ enced by the experimental methods used. For example, grinding the litter applied to the litter addition treatments may have facilitated a more rapid microbial decomposition and subsequent R h response than would have otherwise occurred had whole litter bee n applied to the treatment sites. Due to the decreased surface to volume ratio and increased particle size, it is possible that litterfall under natural field conditions may elicit less of a n immediate microbial response. Additional study of R s rates wit hin the Stoddard Fire Plots using natural litter additions may provide insight into this potential bias.

PAGE 97

97 It has been shown elsewhere that increases in labile soil carbon inputs can induce subsequent rapid and prolonged increases in soil microbial metabolic activity, such that microbial populations mineralize not only the recently added carbon but also older and more recalcitrant soil carbon sources in a process known as priming (Chemidlin Prvost Bour et al., 2010; Kuzyakov, 2010; Blagodatskaya et al., 201 1). In a study of R h rates in France, Chemidlin Prvost Bour et al. (2010) reported evidence of a microbial priming effect within two months of litter additions that persisted for well over a year and resulted in a 32% increase in R s rates. It has been suggested that m icrobial priming may cause significant changes in soil carbon pools as some global climate change and atmospheric nitrogen deposition models forecast increased litterfall in some regions (Hoosbeek, 2004; Kuzyakov, 2010; Sulzman et al., 2012 ). Given the seven month study period reported here it is difficult to say whether or not evidence of a real or prolonged microbial priming effect was observed in the litter additions (Chemidlin Prvost Bour et al., 2010; Sulzman et al., 2012). In the 4 0YR prescribed fire treatments, litter additions recorded a non significant increase (7%) in R s 2 m 2 sec 1 CO 2 m 2 sec 1 ). Given that litter additions in the 40YR sites did not increase R s rates significantly, heterotrophic sources of R s in the 40YR sites may not have been carbon limited. The significantly higher basal area, litter depth, and duff depth in the 40YR sites may have provided ample aboveground carbon inputs and belowground root exudates (and turnover) to support heterotrophic microb ial activity. One possible response to these results would be to suggest a trenching or exclusion study to remove the inputs of recent photosynthate from roots and then reapply fresh litter (Hanson et al., 2000;

PAGE 98

98 Sapronov and Kuzyakov, 2007). Trenching an d exclusion studies however invariably disturb the soil microenvironment often through alterations of soil moisture content due to limitations in lateral diffusion of soil water, increases in dead roots due to severing, and limitations in the lateral diffu sion of soil CO 2 (Kutsch et al., 2010; G. Lokuta, 2012 pers. comm). Given the monthly interval for sampling R s rates in this study, additional short term pulses in soil CO 2 efflux following litter treatments may have gone undetected. Recent research has d emonstrated that soil CO 2 efflux rates are tightly temporally coupled with above and belowground carbon inputs (Stoy et al., 2007). For example, recent research using canopy level 13 C isotope sampling in a loblolly pine forest in North Carolina, USA, dete cted increases in soil CO 2 efflux 13 C fractioning within 3 6 days of the initial canopy exposure (Warren et al., 2012). In another recent example from the Everglades region of Florida, USA, Medvedeff (2012) found that increases in R h in response to experimental ash addit ions were detectable within two days of treatment. Importantly, Medvedeff (2012) also noted that the treatment effects of ash additions were no longer detectable thirty days post treatment. We sugge st that future studies investigating the influence of forest management methods on soil carbon dynamics utilize carbon isotope pulses and repeated short and long term sampling. T hough costly, t hese methods may allow for the differentiation of R a and R h while capturing the highly variable temporal responses of heterotrophic soil microbes following carbon inputs. In addition, such techniques may also provide evidence for microbial priming through the potential identifi cation of the source of soil CO 2

PAGE 99

99 Effect of L itter E xclusion In our study litter exclusion did not reduce R s rates within any prescribed fire interval. The lack of a signif icant treatment effect may have been the result of the study duration or season. Litter exclusion tents were installed in May 2011 after the majority of the deciduous tree and annual grass litter from the previous growing season were shed. Previous resea rch from a similar old field loblolly shortleaf pine successional study site in Mississippi, USA, reported that 90% of the annual deciduous leaf litter load fell between the months of October and December (Hinesley and Nelson, 1991). Given the results of Hinesley and Nelson (1991), it is possible that had the exclusion treatments in our study been installed in September or October of 2010, the exclusion of the litter inputs from the peak 2010 year litterfall period might have influenced R s rates during the sample period. This is supported by the results of Kutsch et al. (2010) who found that models of monthly R h rates for an old growth deciduous forest in Germany were strongly correlated with the litter production from the previous year. Our results are s imilar to those of Garten (2009) who reported that R s rates within litter exclusion treatments in a temperate forest in Tennessee, USA were not significantly different (p > 0.05) from control, while litter additions in the same study resulted in a signifi cant increase (62%) in mean R s rates. In contrast, other studies have found longer term litter exclusion treatments to significantly reduce total R s rates (Bowden et al., 1993; McCarthy and Brown, 2006; Chemidlin Prvost Bour et al., 2010). In an experi mental burning and litter manipulation study in Ohio, USA, McCarthy and Brown (2006) found that litter exclusion and the removal of existing forest floor litter significantly reduced R s rates relative to control. In a field study near Paris, France, Chemi dlin Prvost Bour et al. (2010) reported a significant decrease (25 45%) in R s rates relative to control due to

PAGE 100

100 litter exclusion. Similarly, Bowden et al. (1993) reported a persistent reduction (non significant) in R s rates during a June July study fol lowing litter exclusion during the previous September October in a mixed hardwood forest in Massachusetts, USA. The duration of our study and litter exclusion treatments may have also contributed to the lack of a significant treatment effect from litter ex clusion. Preliminary research on carbon turnover using 14 C isotopic sampling in the 40YR and 3YR Stoddard Research Plots has indicated that the residence time of soil carbon range s from 11 years in the 3YR Stoddard plots to 5.5 years in the 40YR plots (P. Hsieh and K. Robertson, unpublished data). These results and those of others discussed by Sayer (2006) suggest that a 7 month litter exclusion period may not have been long enough to elicit a change in microbial decomposition and subsequent R s rates. D uring our sampling period, existing organic matter in and above the soils in the litter exclusion treatments may have been sufficient to sustain heterotrophic microbial metabolism and subsequent R h rates (Sayer, 2006). A lack of a litter exclusion treatmen t effect could also indicate that in these systems, aboveground inputs such as litterfall may not be as important in driving R s rates as recently fixed photosynthate. Recent research has shown that both recent photosynthate and fine root turnover can be i mportant sources of carbon for forest soil CO 2 efflux in the short term (Epron et al., 2012; Warren et al., 2012). The importance of aboveground litter as a source for soil heterotrophic microbial decomposition may also vary seasonally. For example, Warr en et al. (2012) found evidence suggesting that the role of recent photosynthate in soil CO 2 efflux in a loblolly pine forest in North Carolina, USA declined in importance during the autumn months as other sources of labile carbon

PAGE 101

101 become more important for heterotrophic decomposition during that period Future studies that include prolonged litter exclusion treatments and periodic litter additions that vary seasonally may provide insight into these relationships between aboveground inputs and soil CO 2 effl ux rates. Importance of S oil T emperature In our study soil and ambient temperature explained much of the temporal variation in R s rates, regardless of prescribed fire interval or litter treatment type. Temperature regulates R s rates through the metabolic influence on microbial enzyme activity and is correlated with the seasonal photosynthetic activity of plants (Luo and Zhou, 2006). Our results were consistent with the results of many other studies of upland ecosystems of the southeastern US. For exampl e, Reinke et al. (1981) found that ambient air temperature was highly (R 2 = 0.73) correlated with R s rates in a South Carolina, USA longleaf pine forest, while Fang et al. (1998) found that soil temperature explained > 90% of the observed variability of R s rates in a Florida, USA slash pine plantation. The positive correlations observed in many ecosystems between soil CO 2 efflux rates and soil temperature has led some to raise concerns regarding the future of landscape level and global soil CO 2 efflux rat es under warmer climate conditions (Bond Lamberty and Thomson, 2010). Others have further suggested that as R s rates increase in response to global climate change, the elevated atmospheric CO 2 and global temperature may drive a positive feed back mechanis m resulting in increased R s rates and loss of soil carbon stocks (Rustad et al., 2000). However it is important to consider in these discussions the intricate relationships that exist between aboveground vegetation and belowground soil microbial assemblag es, as experimental manipulations

PAGE 102

102 have found that increases in atmospheric CO 2 concentrations and temperature can result in positive, negative, and or neutral R s responses from belowground soil microbial populations (Wardle et al., 2004; Garten, 2009; Lau and Lennon, 2012). The estimated Q 10 values that describe the response of R s to change s in soil temperature ranged from Q 10 = 1.57 3.40 in our study. These values were similar to those reported in Chapter 2 (this document) following a previous multi yea r study in the Stoddard Fire Plots. In addition, Wang et al. (2006) reported similar Q 10 values for a range of forest types in China (Q 10 = 2.61 3.75), while Samuelson and Whitaker (2012) reported similar Q 10 values (Q 10 = 2.81) following a year long st udy in a natural longleaf pine forest in Alabama, USA. In our study, the highest Q 10 values were observed in the 40YR FRI and the lowest in the 1YR FRI, with litter addition treatments resulting in slightly higher Q 10 values in the 2YR and 40YR FRIs. The results of Bhupinderpal Singh et al. (2003) and Zhou and Zhou (2012) suggest that variability in Q 10 values among treatments and study sites may indicate differences in the relative contributions of R h and R a sources to R s In a few studies (Bhupinderpal Singh et al., 2003; Luan et al., 2011; Zhou and Zhou, 2012) it has been shown through experimental manipulation that heterotrophic sources of R s have higher Q 10 values than autotrophic sources of R s If that is the case, th e n our results indicate that li tter additions may have increased the importance of heterotrophic microbial contributions to total R s in the 2YR and 40YR sites Given that we observed increases in R s rates in all FRIs following litter additions, it makes sense that labile carbon supplem ents would increase heterotrophic microbial activity similar to the results of Medvedeff (2012). It is not clear however, why litter additions in the 1YR treatment, which did increase R s rates, did not result in increased

PAGE 103

103 estimated Q 10 values. In additio n, the results of Bhupinderpal Singh et al. (2003), Luan et al (2011), and Zhou and Zhou (2012) suggest that heterotrophic sources of R s were more significant in the 40YR FRI (litter control Q 10 = 3.16) than in the 1YR (litter control Q 10 = 1.74) and 2YR ( litter control Q 10 = 1.92) FRIs. Previous research using 14 C isotope analysis has shown that soil carbon turnover time in the 40YR treatments is much faster (5.5 years) than in the 3YR Stoddard Fire Plots (11 years) (P. Hsieh, unpublished data ). While th e 3YR Stoddard Fire Plots were not assessed in this study, they are similar in structure and composition to the 2YR sites. The results of P. Hsieh along with those of Hanula et al. (2012) from a study in the Osceola National Forest, suggest that frequent prescribed fire may either directly or indirectly reduce microbial decomposition rates. It must be considered that the use of Q 10 values for partitioning sources of R s is not a practice well established in the literature. Many previous partitioning studie s (Boone et al., 1998; Saiz et al., 2006; Sulzman et al., 2012) have found Q 10 values that contrast with those of Bhupinderpal Singh et al. (2003), Luan et al. (2011), and Zhou and Zhou (2012). For example, following a physical R s partitioning study of a mixed hardwood deciduous forest in Massachusetts, USA, Boone et al., (1998) reported Q 10 values of Q 10 = 4.6 for autotrophic CO 2 efflux, Q 10 = 2.5 for heterotrophic sources of CO 2 efflux, and Q 10 = 3.5 for bulk soil (control). We are hopeful that future s tudies using isotopic sampling or other methods may provide guidance on the disagreement between the results of Q 10 studies as the methods and results of Bhupinderpal Singh et al. (2003), Luan et al (2011), and Zhou and Zhou (2012) could provide a low cost

PAGE 104

104 method of identifying the relative contributions of the sources of soil CO 2 efflux without disturbing integrated soil biogeochemical processes. Importance of S oil M oisture In our study, regardless of prescribed fire treatment or litter manipulation metho d, soil moisture content did not explain a significant amount of the temporal variation in R s rates. These results are consistent with those of other studies in southeastern US upland ecosystems that found little to no correlation between soil moisture co ntent and R s rates (Fang et al., 1998; Gough and Seiler, 2004; Samuelson et al., 2004; Whitaker, 2010). Given that the observed soil moisture content values were not significantly related to the monthly precipitation values, as was expected, we propose th ree possible situations that may have occurred that explain the lack of agreement. First, soil moisture measurements represented a once per month sample of soil moisture conditions within the plot, while total monthly precipitation represented a cumulativ e monthly figure. The lack of correlation may have been due to temporal gaps between the two measurements types. Second, due to large data gaps caused by technical difficulties with the onsite weather station, the monthly precip itation data were recorded at an automated weather station located 30 km from the sample plots. Given the highly heterogeneous nature of precipitation events (Bellon and Austin, 1986), it is possible that the monthly precipitation values for the off site weather station were not r epresentative of the precipitation amounts received at the plots. Finally, we suggest similar to Kutsch et al. (2010), that it is also possible that soil moisture content measurements were biased due to the measurement instrument, as the 5 cm EC 5 probe ( Decagon Systems, Pullman, Washington, USA) used to sample soil moisture content during R s measurements may not have been long enough to measure mineral

PAGE 105

1 05 soil moisture content given the depth of the duff and litter layers in the 40YR plots (4.39 cm) as compa red to the 1YR (1.85 cm) and 2YR treatments (2.63 cm). Recent research from the southeastern US suggests that soil moisture and prolonged drought events can reduce total ecosystem respiration (Bracho et al., 2012) and soil CO 2 efflux (Noormets et al., 201 0) in upland forested ecosystems. Given that our study took place during a prolonged regional drought, the results of Bracho et al. (2012) and Noormets et al. (2010) should be taken into consideration when making comparisons with studies conducted during periods with more or less precipitation. Conclusion s Our results have shown that prolonged p rescribed fire management practices can significantly influence forest soil CO 2 efflux rates in the loblolly pine shortleaf pine old field forests of North Florid a, USA Frequent burning reduces soil CO 2 efflux rates in the study area relative to fire exclusion, with annual burning resulting in lower monthly mean soil CO 2 efflux rates than biennial burning. Our results also found that soil CO 2 efflux rates can in crease for a period of several months following one time litter additions, with the greatest increases in the annually and biennially burned sites. At the same time, soil CO 2 efflux rates in both frequently burned and long fire excluded sites do not appea r sensitive to short term reductions in leaf litter inputs. Even though a positive soil CO 2 efflux response was detected following experimental litter additions, it remains to be seen whether litter represents a significant, dominant, or seasonal source o f labile carbon for soil heterotrophic microbial respiration, as previous studies in other systems have reported the importance of fine root turnover and root exudates in supplying heterotrophic microbial respiration. Our results provided some evidence to suggest that the role of leaf litter in soil CO 2 efflux differs between sites managed with various

PAGE 106

106 prescribed fire management regimes. Future research from a soil microbial ecology perspective may allow for a better understanding of how prolonged prescri bed fire management regimes, and vegetative above and belowground inputs shape soil bacterial and fungal populations responsible for heterotrophic soil CO 2 efflux.

PAGE 107

107 Table 3 1. Plot level variables investigated for their i nfluence on soil CO 2 efflux rates at the Tall Timbers Research Station, FL Parameter category Plot variable Abbreviation Measured Measurement location Microclimate Soil temperature T s (C) 3x daily 5 15 cm of collar Soil moisture content M s (m 3 /m 3 ) 3x daily 5 15 cm of collar Veget ation Basal area BA (m 2 ha 1 ) Winter 2011 15 m radius circular plot from center collar Pine basal area PBA (m 2 ha 1 ) Winter 2011 15 m radius circular plot from center collar Hardwood basal area HWBA (m 2 ha 1 ) Winter 2011 15 m radius circular plot fro m center collar Stand density TPH (trees ha 1 ) Winter 2011 15 m radius circular plot from center collar Forest floor Duff depth Duff (cm) Spring 2011 Avg. of three measurements within 30 cm of collar Litter depth Litter (cm) Spring 2011 Avg. of thre e measurements within 30 cm of collar Total duff and litter depth DL (cm) Spring 2011 Avg. of three measurements within 30 cm of collar Weather Total precipitation Precip (cm) Monthly Quincy, FL FAWNS station Mean air temperature (2 m) Temp (C) Month ly Quincy, FL FAWNS station Palmer drought severity index PDSI Monthly Northwest Florida regional estimate from NOAA NCDC

PAGE 108

108 Table 3 2. Mean forest characteristics per prescribed fire treatment type at the Tall Timbers Research Station, FL FRI Year Tree s (ha) Hardwood basal area (m 2 ha 1 ) Pine basal area (m 2 ha 1 ) Total basal area (m 2 ha 1 ) Duff depth (cm) Litter depth (cm) Annual litterfall (t ha 1 yr 1 ) 1YR 2011 282.93 (64.83) b 3.87 (6.28) b 7.92 (3.68) a 11.79 (7.22) b 0.08 (0.06) c 1.77 (0.91) c 4.63 (1.53) a 2YR 2011 400.81 (344.87) b 6.30 (4.28) ab 9.16 (6.14) a 15.45 (2.15) b 0.46 (0.41) b 2.17 (0.79) b 5.42 (0.69) a 40YR 2011 1716.41 (681.42) a 15.73 (3.59) a 21.99 (11.22) a 37.72 (8.36) a 1.58 (0.55) a 2.81 (0.58) a 5.47 (1.37) a ; Tukey 1953). Litterfall rates provided by K. Robertson (unpublished data)

PAGE 109

109 Table 3 3. Results of the repeat ed measures ANOVA for soil CO 2 efflux (R s ), soil temperature (T s ), and soil moisture content (M s ) means at the Tall Timbers Research Station, FL R s T s M s Term df F p df F p df F p FRI 2 24.20 < 0.0001 2 6.80 0.0016 2 43.06 < 0.0001 Litte r 2 11.29 < 0.0001 2 0.03 0.9677 2 7.00 0.0014 Time 6 69.89 < 0.0001 5 918.45 < 0.0001 5 59.86 < 0.0001 FRI*Time 12 3.92 < 0.0001 10 14.73 < 0.0001 10 2.18 0.0245 Litter*Time 12 0.49 0.9150 10 0.08 1.0000 10 1.20 0.0402 FRI*Litter 4 1.36 0.2514 4 0.01 1.0000 4 1.51 0.2039 For each month, daily measurement s per soil collar were averaged and the three soil collar means per litter treatment type were then averaged to produce a plot l e vel mean value for each month.

PAGE 110

110 Table 3 4. Mean soil CO 2 efflux rate, soil temperature, and soil moisture content for the entire study period by prescribed fire treatment at the Tall Timbers Research Station, FL Fire return interval (FRI) Mean R s 2 m 2 sec 1 ) Mean T s ( C) Mean M s (m 3 /m 3 ) 1YR 3.44 (2.57) c 21.21 (7.33) a 0.40 (0.08) a 2YR 4.50 (3.45) b 20.88 (6.35) ab 0.13 (0.07) a 40YR 5.15 (3.92) a 20.46 (4.64) b 0.08 (0.06) b Values are means with standard deviation in parentheses. Means are for fire return intervals with litter treatments grouped. Letters indicate significant differences among month from June December 2011. Table 3 5. Mean soil C O 2 efflux rate, soil temperature, and soil moisture content for the entire study period by litter treatment type at the Tall Timbers Research Station, FL Litter treatment type Mean R s 2 m 2 sec 1 ) Mean T s ( C) Mean M s (m 3 /m 3 ) Addition 5.05 (3.93) a 20.83 (6.24) a 0.12 (0.08) a Exclusion 4.10 (3.18) b 20.83 (6.20) a 0.10 (0.07) b Control 3.94 (3.01) b 20.89 (6.22) a 0.13 (0.08) a Values are means with standard deviation i n parentheses. Means are for litter treatment types with prescribed fire return interval (FRI) grouped. Letters indicate three times daily once per month from June Dece mber 2011.

PAGE 111

111 Table 3 6. Results of the repeated measures ANOVA for soil CO 2 efflux (R s ), soil temperature (T s ), and soil moisture content (M s ) means within prescribed fire treatments at the Tall Timbers Research Station, FL R s T s M s FRI Term df F p df F p df F p 1YR Litter 2 6.65 0.0031 2 0.01 0.9885 2 6.39 0.0042 Time 6 17.28 < 0.0001 5 212.54 < 0.0001 5 24.94 < 0.0001 Litter*Time 12 0.34 0.9770 10 0.03 1.0000 10 1.52 0.1738 2YR Litter 2 7.42 0.0017 2 0.02 0.9846 2 0.49 0.6153 Time 6 20.28 < 0.0001 5 459.17 < 0.0001 5 22.66 < 0.0001 Litter*Time 12 0.54 0.8740 10 0.10 0.9997 10 0.36 0.9544 40YR Litter 2 0.94 0.3974 2 0.03 0.9961 2 2.69 0.0814 Time 6 36.31 < 0.0001 5 777.91 < 0.0001 5 14.07 < 0.0001 Litter*Time 12 0.18 0.9988 10 0.10 0.9997 10 1.30 0.2681 For each month, daily measurement s per soil collar were averaged and the three soil collar means pe r litter treatment type were then averaged to produce a plot le vel mean value for each month.

PAGE 112

112 Table 3 7. Soil CO 2 efflux, soil temperature, and soil moisture content means by litter treatment type and fire return interval for the Tall Timbers Research Station, FL Fire return interval (FRI) Litter treatment Mean R s 2 m 2 sec 1 ) Mean T s ( C) Mean M s (m 3 /m 3 ) 1YR Litter 4.21 (3.05) a 21.12 (7.38) a 0.14 (0.08) ab 1YR Exclusion 3.06 (2.07) b 21.22 (7.34) a 0.12 (0.08) b 1YR Control 3. 05 (2.34) b 21.28 (7.34) a 0.16 (0.08) a 2YR Litter 5.45 (4.55) a 20.89 (6.41) a 0.13 (0.07) a 2YR Exclusion 4.38 (2.72) ab 20.84 (6.26) a 0.12 (0.07) a 2YR Control 3.68 (2.47) b 20.92 (6.44) a 0.13 (0.07) a 40YR Litter 5.49 (3.9 3) a 20.48 (4.64) a 0.09 (0.06) a 40YR Exclusion 4.86 (4.12) a 20.42 (4.74) a 0.06 (0.04) a 40YR Control 5.09 (3.68) a 20.47 (4.56) a 0.07 (0.06) a Values are means with standard deviation in parentheses. Means are for litter treatm ent type by fire return interval. Letters indicate significant differences among litter recorded three times daily once per month from June December 2011 for three prescribed fire treatment intervals.

PAGE 113

113 Table 3 8. Linear regression models of the relationships between soil CO 2 efflux rates and soil temperature by fire return interval and litter treatment type FRI Litter treatment Model and estimates F R 2 p 1YR Litter R s = 0. 1074 + 0.1768*T s 9.77 0.38 0.0065 1YR Exclusion R s = 0.1272 + 0.1320* T s 14.03 0.47 0.0018 1YR Control R s = 0.5696 + 0.1547* T s 17.41 0.52 0.0007 2YR Litter R s = 2.4001 + 0.3508*T s 18.80 0.54 0.0005 2YR Exclusion R s = 1.1155 + 0.2418*T s 22.60 0.59 0.0002 2YR Control R s = 1.1360 + 0.2089*T s 24.47 0.61 0.0001 40YR Litter R s = 4.2070 + 0.4523*T s 9.78 0.38 0.0065 40YR Exclusion R s = 3.7433 + 0.3982*T s 11.23 0.41 0.0041 40YR Control R s = 3.8450 + 0.4130*T s 10.94 0.41 0.0044 Model data are mean monthly measurements from June December 2011 taken at the Tall Timbers Research Station near Tallahassee, Florida, USA. R s is soil CO 2 2 m 2 sec 1 ), T s is soil temperature (C).

PAGE 114

114 Table 3 9. Non linear exponential models of the rela tionships between soil CO 2 efflux rates (R s ) and soil temperature by fire return interval and litter treatment type r Data are results of non linear exponential models (R s ) of soil CO 2 efflux rate (R s ) ( mol CO 2 m 2 sec 1 ) responses to soil temperature (T s ). Data are presented by prescribed fire return interval (FRI) and litter manipulation treatment type. Coefficients were estimated using statistical software SAS JMP 9.0. Q 10 was calculated using the exponential equation Q 10 = e linear model. Fire return interval (FRI) Treatment Q 10 R 2 p 1YR Litter 1.4158 0.0450 1.57 0.36 0.0086 1YR Exclusion 0.91 92 0.0476 1.61 0.43 0.0030 1YR Control 0.7791 0.0556 1.74 0.49 0.0013 2YR Litter 0.9544 0.0741 2.10 0.52 0.0007 2YR Exclusion 1.0081 0.0621 1.86 0.55 0.0004 2YR Control 0.7652 0.0653 1.92 0.58 0.0003 40YR Litter 0.3507 0.1224 3.40 0.44 0.0028 40YR Ex clusion 0.3257 0.1197 3.31 0.46 0.0020 40YR Control 0.3819 0.1150 3.16 0.45 0.0023

PAGE 115

115 Table 3 10. Linear regression of soil temperature and monthly mean ambient air tempera ture by litter treatment type and fire return interval Fire return interval (FRI) Litter treatment Relationship R 2 F p 1YR Addition Pos 0.92 188.12 < 0.0001 1YR Exclusion Pos 0.92 180.88 < 0.0001 1YR Control Pos 0.93 197.22 < 0.0001 2YR Addition Pos 0. 94 257.95 < 0.0001 2YR Exclusion Pos 0.95 298.56 < 0.0001 2YR Control Pos 0.94 251.76 < 0.0001 40YR Addition Pos 0.95 292.70 < 0.0001 40YR Exclusion Pos 0.94 252.98 < 0.0001 40YR Control Pos 0.95 280.16 < 0.0001 Soil temperature data (T s ) (C) were r ecorded in sample plots at Tall Timbers Research Station, Florida, USA. Monthly mean ambient air temperature (M Temp) (C) data were means from hourly 2 m measurements recorded at the Florida Automated Weather Network Station (FAWNS) in nearby Quincy, FL.

PAGE 116

116 Table 3 11. Linear regression of the relationships between soil CO 2 efflux rates and soil moisture content by litter treatment type and fire return interval FRI Litter treatment Model and estimates F R 2 p 1YR Litter R s = 2.6386 + 10.1666*M s 2.38 0.1 3 0.1422 1YR Exclusion R s = 3.7252 6.4439*M s 1.13 0.07 0.3039 1YR Control R s = 2.6459 + 2.0158*M s 0.13 0.01 0.7357 2YR Litter R s = 3.9793 + 8.7733*M s 0.72 0.04 0.7154 2YR Exclusion R s = 4.7924 4.6506*M s 0.23 0.01 0.6381 2YR Control R s = 3.5785 0 .2566*M s 0.00 0.00 0.9732 40YR Litter R s = 3.6307 + 21.6083*M s 1.63 0.09 0.2200 40YR Exclusion R s = 4.4822 + 3.2570*M s 0.02 0.00 0.8887 40YR Control R s = 3.1391 + 23.5037*M s 3.59 0.18 0.0765 Model data are mean monthly measurements from June December 2 011 taken at the Tall Timbers Research Station near Tallahassee, Florida, USA. R s is soil CO 2 2 m 2 sec 1 ), M s is soil moisture content (m 3 /m 3 ).

PAGE 117

117 Table 3 12. Linear regression of the relationships between soil CO 2 efflux rates and m onthly precipitation by litter treatment type and fire return interval FRI Litter treatment Model and estimates F R 2 p 1YR Litter R s = 1.0919 + 0.2961*Precip 19.94 0.51 0.0003 1YR Exclusion R s = 0.6766 + 0.2268*Precip 16.34 0.46 0.0007 1YR Contro l R s = 0.4195 + 0.2507*Precip 25.98 0.58 < 0.0001 2YR Litter R s = 0.6896 + 0.4484*Precip 24.61 0.56 < 0.0001 2YR Exclusion R s = 1.2985 + 0.2936*Precip 19.41 0.51 0.0003 2YR Control R s = 0.8007 + 0.2743*Precip 20.29 0.52 0.0002 40YR Litter R s = 0 .4857 + 0.5698*Precip 52.57 0.73 < 0.0001 40YR Exclusion R s = 0.4015 + 0.5011*Precip 53.82 0.74 < 0.0001 40YR Control R s = 0.2653 + 0.5111*Precip 57.08 0.75 < 0.0001 Model data are mean monthly measurements from June December 2011 taken at the Tall Ti mbers Research Station near Tallahassee, Florida, USA. R s is soil CO 2 2 m 2 sec 1 ), Precip is monthly total precipitation for the region from measurements recorded at the Florida Automated Weather Network Station (FAWNS) in nearby Quincy, FL.

PAGE 118

118 Table 3 13. Linear regression of soil moisture content and monthly precipitation by litter treatment type and fire return interval Fire return interval (FRI) Litter treatment Relationship R 2 F p 1YR Addition Pos 0.14 2.62 0.1250 1YR Exclusion Neg 0.07 1.20 0.2892 1YR Control Pos 0.08 1.47 0.2428 2YR Addition Pos 0.15 2.92 0.1070 2YR Exclusion Pos 0.02 0.41 0.5324 2YR Control Pos 0.03 0.43 0.5232 40YR Addition Pos 0.10 1.83 0.1946 40YR Exclusion Neg 0.02 0.27 0.6092 40YR Control Pos 0.23 4.89 0.0420 Model soil moisture content (M s ) (m 3 /m 3 ) data are mean m onthly measurements from June December 2011 taken at the Tall Timbers Research Station near Tallahassee, Florida, USA. R s is soil CO 2 2 m 2 sec 1 ), Precip is monthly total precipitation for the region from measurements recorded at the Florida Automated Weather Network Station (FAWNS) in nearby Quincy, FL.

PAGE 119

119 Table 3 14. Linear regression of soil temperature and monthly precipitation by litter treatment type and fire return interval Fire return interval (FRI) Litter treatment Relations hip R 2 F p 1YR Addition Pos 0.60 24.00 0.0002 1YR Exclusion Pos 0.58 21.92 0.0003 1YR Control Pos 0.58 22.46 0.0002 2YR Addition Pos 0.63 27.13 < 0.0001 2YR Exclusion Pos 0.62 25.88 0.0001 2YR Control Pos 0.61 24.80 0.0001 40YR Additi on Pos 0.63 27.70 < 0.0001 40YR Exclusion Pos 0.62 25.91 0.0001 40YR Control Pos 0.63 27.29 < 0.0001 Soil temperature data (T s ) (C) data are mean monthly measurements from June December 2011 taken at the Tall Timbers Research Station near Tallahasse e, Florida, USA. Precipitation is monthly total precipitation for the region from measurements recorded at the Florida Automated Weather Network Station (FAWNS) in nearby Quincy, FL.

PAGE 120

120 Figure 3 1. Map of the study area at the Tall Timbers Research Sta tion in Leon County, Florida, USA. Map produced by David Godwin.

PAGE 121

121 Figure 3 2. Ground (left) and aerial (right) images of three of the soil CO 2 efflux sampling plots located within the Tall Timbers Research Station in Leon County, Florida, USA. The top images show an annual burn frequency site (1YR), the middle images a two year burn frequency site (2YR), and the bottom image a site unburned since 1966. Ground images original to the author. Ground photographs courtesy of David Godwin. Aerial image s courtesy of Microsoft Bing Maps.

PAGE 122

122 Figure 3 3. Photograph of a 20 cm soil CO 2 efflux sample collar and 0.16 m 2 wood treatment box (top) and litter exclusion enclosure (bottom) at the Tall Timbers Research Station, Florida, USA. Photograp hs courtesy of David Godwin.

PAGE 123

123 Figure 3 4. Photograph of the LICOR Biosciences LI 8100 soil CO 2 efflux sampling instrument with 20 cm survey chamber and soil moisture and temperature probes. Photograph courtesy of David Godwin.

PAGE 124

124 Figure 3 5. Plot of seven months of 2 m air temperature records and precipitation for the year 2011 from the Florida Automated Weather Network (FAWN) site at Quincy, Florida, approximately 30 km from Tall Timbers Research Station, Florida, USA.

PAGE 125

125 Figure 3 6. Plot of seven months of monthly Palmer Drought Severity Index (PDSI) values for the year 2011 from the National Oceanic and Atmospheric Administration (NOAA) National Climatic Data Center (NCDC). All scores below zero represent drought conditions for the region.

PAGE 126

126 F igure 3 7. Monthly mean soil CO 2 efflux rates (R s ) ( mol CO 2 m 2 sec 1 ), soil temperature (T s ) ( C ), and soil moisture content (M s ) (m 3 /m 3 ) by litter treatment type (litter addition, litter exclusion, and control). Points indicate monthly means of all st udy data with FRI treatment type ignored. Equipment problems resulted in no M s data collected during the month of June and no T s data collected during the month of August.

PAGE 127

127 Figure 3 8. Monthly mean soil CO 2 efflux rates (R s ) ( mol CO 2 m 2 sec 1 ), soil temperature (T s ) ( C ), and soil moisture content (M s ) (m 3 /m 3 ) prescribed fire management type (1YR, 2YR, and 40YR). Points indicate monthly means of all study data with litter treatment type ignored. Equipment problems resulted in no M s data collected d uring the month of June and no T s data collected during the month of August.

PAGE 128

128 Figure 3 9. Monthly mean soil CO 2 efflux rates (R s ) ( mol CO 2 m 2 sec 1 ) by litter (litter addition, litter exclusion, and control) and fire (1YR, 2YR, and 40YR) treatment type. Points indicate soil respiration rate monthly averages for the study period June December 2011.

PAGE 129

129 Figure 3 10. Overall mean soil CO 2 efflux rates (R s ) ( mol CO 2 m 2 sec 1 ) by litter manipulation treatment (litter addition, exclusion, and control) within fire return interval treatment (1YR, 2YR, and 40YR). Letters indicate significant differences among litter treatments w test = 0.05). Data are means for the study period June December 2011.

PAGE 130

130 Figure 3 11. Monthly mean soil moisture content (M s ) (m 3 /m 3 ) by litter (litter addition, litter exclusion, and control) and fire (1YR, 2Y R, and 40YR) treatment type. Points indicate soil moisture content monthly means for the study period June December 2011. Equipment problems resulted in no data collected during the month of June.

PAGE 131

131 Figure 3 12. Overall mean soil moisture content (M s ) (m 3 /m 3 ) by litter manipulation treatment (litter addition, exclusion, and control) within fire return interval treatment (1YR, 2YR, and 40YR). Letters indicate significant differences = 0.05). Data are means for the study period June December 2011.

PAGE 132

132 Figure 3 13. Monthly mean soil temperature (T s ) ( C ) by litter (litter addition, litter exclusion, and control) and fire (1YR, 2YR, and 40YR) treatment type. Points indicate soil temperature monthly averages for the study period June December 2011. Equipment problems resulted in no data collected during the month of August.

PAGE 133

133 Figure 3 14. Overall mean soil temperature (T s ) ( C) by litter manipulation treatment (litter additio n, exclusion, and control) within fire return interval treatment (1YR, 2YR, and 40YR). Letters indicate significant differences among litter = 0.05). Data are means for the study period June December 2011.

PAGE 134

134 Figure 3 15. Linear regression of the relationships between monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and soil temperature (T s ) ( C) for three litter treatment ty pes within the 1YR prescribed fire interval at the Tall Timbers Research Station near Tallahassee, Florida, USA. Each point represents monthly mean values per sample plot.

PAGE 135

135 Figure 3 16. Linear regression of the relationships between monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and soil temperature (T s ) ( C) for three litter treatment types within the 2YR prescribed fire interval at the Tall Timbers Research Station near Tallahassee, Florida, USA. Each point represents monthly mean values per sample plot.

PAGE 136

136 Figure 3 17. Linear regression of the relationships between monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and soil temperature (T s ) ( C) for three litter treatment types within the 40YR prescribed fire interval at the Tall Timbers Rese arch Station near Tallahassee, Florida, USA. Each point represents monthly mean values per sample plot.

PAGE 137

137 Figure 3 18. The relationship between monthly mean soil CO 2 2 m 2 sec 1 ) ( R s ) and monthly mean soil temperature (C) (T s ) as modeled using an exponential equation (Equation 3 2). Data presented are from the 1YR prescribed fire treatment interval litter treatment types. Each point represents monthly mean values per sample plot.

PAGE 138

138 Figure 3 19. The relationship between monthly mean soil CO 2 efflux rate 2 m 2 sec 1 ) ( R s ) and monthly mean soil temperature (C) (T s ) as modeled using an exponential equation (Equation 3 2). Data presented are from the 2YR prescribed fire treatment interval litter treatment types. Each point represents monthly mean values per sample plot.

PAGE 139

139 Figure 3 20. The relationship between monthly mean soil CO 2 2 m 2 sec 1 ) ( R s ) and monthly mean soil temperature (C) (T s ) as modeled using an exponential equation (Equation 3 2). Data presented are from t he 40YR prescribed fire treatment interval litter treatment types. Each point represents monthly mean values per sample plot.

PAGE 140

140 CHAPTER 4 THE INFLUENCE OF PRE SCRIBED FIRE AND UND ERSTORY FUELS MASTICATION ON SOIL CO2 EFFLUX RATES IN TWO NORTH FLORIDA FLA TWOODS FORESTS Background It is important to understand the implications of forest management practices on soil carbon dynamics as forests and forest soils play significant roles in global carbon cycles. In temperate forest ecosystems, approxi mately 50 60% of ecosystem carbon is found within the soils, with soil CO 2 efflux (R s ) comprising 50 60% of total ecosystem carbon budgets (Raich and Schlesinger, 1992 ; Lal, 2005; Noormets et al. 2010) One method for assessing how management affects fore st carbon dynamics is the measurement of soil CO 2 efflux rates. A variety of forest management activities, including prescribed fire and mechanical fuels masti cation, have been shown to significantly influence soil CO 2 efflux rates in the w estern United S tates (US) yet these relationships are not well known in southeastern US forests (Concilio et al., 2005; Kobziar, 2007; Ryu et al., 2009 ) Prescribed fire is one of the most prevalent forest management tools employed in the s outheastern US with over 2.4 million ha burned in 2011 and mechanical fuels mastication treatments are becoming more common in the region as concerns o ver wildfire in the wildland urban interface grow (Agee and Skinner, 2005 ; Waldrop and Goodrick, 2012 ) This study seeks to understan d the influence of prescribed fire management regimes and mechanical fuels mastication treatments on soil CO 2 efflux rates in mature pine flatwoods forests of n orth c entral Florida, USA. Flatwoods are the most common forested ecosystem in Florida totaling over 5.2 million ha (Myers and Ewel, 1990) Given the prevalence of this forest type, public and private interest in the effect of forest management activities on forest carbon dynamics

PAGE 141

141 are expected to be significant (Law and Harmon, 2011) With the poss ibility of future federal land management goals including carbon sequestration, understanding the effect of management regimes on carbon dynamics is critical (Exec. Order No. 13,513, 2009). In flatwoods managed for conservation, prescribed fire is one of the most frequently utilized management tools for maintaining ecosystem composition and structure and reducing wildfire risk (Outcalt and Wade, 1999) For wildfire risk reduction in flatwoods forests, prescribed fire is typically applied on a 3 5 year int erval to remove the accumulation of saw palmetto ( Serenoa repens (Bartr.) Small) and other understory vegetative fuels that tend to drive fire behavior in these systems (Brose and Wade, 2002) Given the importance of prescribed fire in these forests and t he potential ecological and economic benefits of carbon credits for carbon sequestration, it is important to understand the influence of these management fires on soil CO 2 efflux rates. As the extent of the wildland urban interface (WUI) has expanded in th e later part of the 20 th and early 21 st century, so has the social and political pressure to reduce the risk of property damage in the interface from wildfires (Vince et al., 2005) There are three primary methods which have been employed to alter forest structure in efforts to reduce such risk: prescribed fire, mechanical mastication and a combination of mechanical mastication followed by prescribed fire (Agee and Skinner, 2005; Kobziar and Stephens, 2006; Hurteau and North, 2009) Mechanical fuels masti cation is used to reduce understory fuel heights, there by increasing the height to live crowns, which has been shown to reduce fire behavior in both western and eastern US forests (Agee and Skinner, 2005; Glitzenstein et al., 2006; Kobziar et al., 2009) In many WUI areas

PAGE 142

142 in the southeastern US, prescribed fire has become difficult for land managers to implement due to concerns from adjacent and nearby landowners over smoke and wildfire risk or because of prolonged fire suppression and subsequent hazardous fuel load accumulations (Miller and Wade, 2003; Long et al., 2004) The use of mechanical fuels treatments with or without prescribed fire has increased in recent years in Florida, USA, as land managers seek to maintain and restore forest structure in ar eas where the implementation of prescribed fire has proven difficult Pre burning mechanical techniques are also applied to alter the arrangement of vegetative fuels to decrease fire intensity and severity upon the subsequent reintroduction of prescribed fire (Menges and Gordon, 2010) As the implementation of these mechanical treatments becomes more widespread it will become even more important to understand their influence on forest carbon. Previous studies of mechanical fuels mastication have shown t hat treatments can significantly alter soil CO 2 efflux rates (Kobziar and Stephens, 2006) as well as influence soil environmental factors such as soil temperature and soil moisture content; factors that are known to drive soil CO 2 efflux rates in some ecos ystems (Concilio et al., 2005; Kobziar and Stephens, 2006; Xu et al., 2011) Mechanical fuels mastication treatments and prescribed fire can influence soil CO 2 efflux rates by altering soil and environmental physical, chemical, and abiotic factors that af fect the sources of heterotrophic and autotrophic R s F or example, fire has been shown to alter forest floor litter and duff loads carbon and nitrogen pools, soil temperature, pH, and microbial activity in multiple ecosystems ( Neary, 1999; Debano, 2000 ). In addition, mechanical fuels mastication treatments have been shown to influence forest floor litter and duff loads and average soil temperature and moisture

PAGE 143

143 content which can also influence heterotrophic and autotrophic sources of R s (Luo and Zhou, 200 6; Kobziar, 2007). Finally, both treatments have clear impacts on understory forest vegetation through physical mastication or damage, combustion, injury, or competitive release that can alter vegetative activity and belowground carbon allocation. Numerou s studies have investigated carbon dynamics in flatwoods and similar commercial slash pine forests of the southeastern US, however none known have specifically addressed the effects of mechanical fuels mastication treatments and prescribed fire management regimes on soil CO 2 efflux rates ( Ewel et al., 1987a; Ewel et al.,1987b; Fang et al., 1998; Clark et al., 2004; Powell et al., 2008; Meigs et al., 2009; Lavoie et al., 2010 ; Bracho et al., 2012 ) By investigating prescribed fire, mechanical fuels masticat ion, and mechanical fuels mastication fo llowed by prescribed fire in the context of two flatwoods ecosystems managed for conserv ation and multiple use purposes, this study sought to address the following research questions: (1) How d o prescribed fire and u nderstory fuels mastication treatments influence monthly, seasonal, and annual soil CO 2 efflux rates, and, (2) How do prescribed fire and mastication treatments a ffect forest conditions that will likely influence long term site level soil CO 2 efflux rates and soil carbon dynamics? For managers and researchers alike, this study provides insight into linkages between forest management, soil carbon storage and flux, and the physical and biotic variables influencing those fluxes that are likely to be influence d by global climate change. Methods Study A rea s The first study sites were located within the 80,000 ha United States Forest Service (USFS) Osceola National Forest ( Osceola ) in Columbia County, FL, USA

PAGE 144

144 approximately 20 km from the town of Lake City (3 Figure 4 1). The area is within the Gulf Coastal Plain region and is generally flat with little to no perceptible slope. The study sites are located approximately 44 m above sea level. Average annual precipitation was 132 cm with t he majority falling during the summer months of June, July and August (National Climate Data Center 2009). Mean maximum and minimum temperatures for January and July for the study area from long term records were 18.9 C and 6.1 C for January and 32.7 C and 21.7 C for July (National Climate Data Center 2009). Soils within the site are generally poorly drained sandy, siliceous, hyperthermic Ultic Alaquods of the Mascotte and Olustee series ( Natural Resource Conservation Service (NRCS) Soil Survey Geogra phic Database (SSURGO) ) Vegetation across all sites consist ed of an overstory mixture of naturally regenerated slash pine ( Pinus elliottii Engelm) and longleaf pine ( P palustris P. Mill) and an understory composed of saw palmetto ( Serenoa repens ( W. Bar tram) Small ), gallberry ( Ilex glabra ), and deerberry ( Vaccinium stamineum ) shrubs (Myers and Ewel 1990) Across the s tudy area stand age averaged 80 years (Osceola National Forest staff pers. comm.). Prior to the start of the study, all plots ha d been unburned for at least 11 years (Jesse Kreye, pers. comm.). The second study site was located within the 840 ha University of Florida Austin Cary Memorial Forest (ACMF) in Alachua County Florida, USA approximately 14 km from the city of Gainesville (29 082 Fig ure 4 1). The site is approximately 44 m a.s.l. and generally flat with no perceptible slope. Average annual precipitation is 123 cm with the majority falling during the summer months National Climate Data Center 2009 ) Mean maximum and minimum temperatures for January

PAGE 145

145 and July for the study area from long term records are 19 C and 6.1 C for January and 36. 6 C and 22.9 C for July (National Climate Data Center 2004). Soils within the site are generally poorly drained sandy, siliceous, hyperthermic Ultic Alaquods of the Pomona series ( NRCS SSURGO ). Vegetation across all sites consists of an overstory mixtu re of naturally regenerated slash pine and longleaf pine and an understory composed of saw palmetto, gallberry, and deerberry shrubs (Myers and Ewel 1990) Across the s tudy area stand age averaged 80 years with an average height of 24 m (Daniel Schultz pers. comm.). Sampling The Osceola National Forest study consisted of twelve sample plots representing four treatment types: prescribed fire (burn), mechanical fuel mastication (mow), mechanical mastication + prescribed fire (mow+burn), and unburned control (control) ( Figure 4 2). Plots were established in three 2 ha experimental treatment blocks, with each block containing a representative plot of each treatment. Three s ampling plots were randomly located within each treatment type in each block. Mechanical fuel mastication in the mow and mow+burn plots took place during the summer of 2010 P rescribed burning in the mow and mow+burn plots took place in February 2011, with two blocks burned on one day and one block burned the next day Blocks were burned by hand using low in tensity strip head fires. Air temperature on the day s of burning ranged from 17 24C and relative humidity (RH) ranged from 47 62%. Regional Keetch Byram Drought Index at the time of burning was 107. The ACMF study consisted of six sample plots represent ing two treatment types: three year prescribed fire interval (3YR), and fire exclusion (40YR) ( Figure 4 3). The study was arranged in a pseudo replicate sampling design due to the limited availability

PAGE 146

146 of treatments, with three sample plots randomly establ ished within each 20 ha treatment type Fire had been excluded from the 40YR treatment for at least forty years while t he nearby 3YR treatment area had been maintained in a three year dormant season prescribed fire interval for at least twelve years prior to the study and had been frequently burned prior to that (Daniel Schultz pers. comm.). Plots in the 3YR treatment were last prescribe d burned in February o f 2009 using strip head fires. Further descriptions of the conditions and fire behavior during th e day of prescribed burning in 2009 were not available. Field M easurements Soil CO 2 efflux sample plots were established in the early winter of 2009 at the ACMF and in the early winter of 2010 at the Osceola study sites Each sample plot consi sted of nine permanently installed 20 cm diameter independent PVC collars arranged in a 3 x 3 grid with 5 m separation following Kobziar and Stephens (2006). PVC sampling collars were constructed of Schedule 30 white 20 cm diameter pipe cut to 10 cm lengt hs and beveled along one edge. Collars were inserted beveled edge down into the soil or duff to a depth of approximately 8 cm using a rubber mallet. All collars were installed at least four weeks prior to the start of sampling to allow any soil disturban ce from installation to normalize. During the course of study, any vegetative growth within the sample collars was clipped and removed prior to R s measurement. A LI COR Biosciences LI 8100 Automated Soil CO 2 Flux System attached to a 20 cm survey chamber was used to measure soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) at each collar (LI COR Bi osciences, Inc., Lincoln, NE). Concurrently with soil CO 2 efflux measurements, soil temperature at 10 cm depth (C) ( T s ) and soil volumetric moisture content (m 3 x m 3 ) ( M s ) at 5 cm depth were recorded onboar d the LI 8100. Soil

PAGE 147

147 temperature ( T s ) was measured using an Omega 8831 type E T Handle temperature probe, while soil moisture content ( M s ) was measured using a Decagon Systems EC 5 soil moisture probe (Omega Inc., Stamford, CT; Decagon Systems Inc., Pullma n, WA). Both probes were inserted into the soil at random azimuths approximately 5 15 cm from the collar and remained undisturbed in the soil during the 120 second R s measurement period. To assess temporal and seasonal variations in R s rates, T s and M s collars were sampled monthly over the course of one (ACMF) or two days (Osceola) on an approximately four week rotation. To account for diurnal variations measurements at the ACMF plots were taken three times per day between 0800 and 1900 local time an d measurements at the Osceola plots were taken twice per day between 0800 and 1700 local time. M onthly mean 2 m ambient air temperature ( C ) records (M Temp) from hourly meteorological observations at the Olustee, Florida remote automated weather station ( RAWS) (approximately 7 km from the Osceola sites) and the Austin Cary Memorial Forest AmeriFlux tower (located within the ACMF 3YR treatment stand) were recorded throughout the study periods S oil temperature ( C ) at 10 cm soil depth measured hourly at th e nearby Florida Automated Weather Network (FAWN) stations in Macclenny, Florida (approximately 30 km from the Osceola study sites ) and Putnam Hall, Florida (approximately 21 km from the ACMF study sites ) were recorded throughout the study periods. The FA WN station hourly 10 cm soil temperature measurements were used as input for the total monthly and annual carbon flux models per site Monthly total precipitation (cm) measurements from the Olustee RAWS station and Austin Cary Forest staff meteorological records were also recorded and assessed

PAGE 148

148 for their influence on R s rate variability. Monthly sampling of R s T s and M s at the ACMF study site began in March 2010 and concluded in June 2011, while monthly sampling at the Osceola study site began in March 2 011 and concluded in March 2012. At the ACMF study site, monthly R s measurements were not collected during the months of August and December of 2010 due to hazardous weather conditions while measurements collected during September 2010 were discarded due to sampling equipment error. At the Osceola study site, monthly measurements were not collected during the month of September 2011 due to hazardous weather conditions. Errors encountered with the soil temperature probe resulted in T s measurement gaps dur ing the months of July 2010 and January 2011 at the ACMF study site and July and August of 2011 at the Osceola study site. Errors encountered with the soil moisture probe resulted in M s measurement gaps during the month of June 2010 at the ACMF study site Recorded soil moisture content values less than 0.00, and soil temperature measurements greater than 40 C were excluded from the analyses, as they resulted from equipment malfunction. Plot characteristics and vegetative sampling were conducted in the wi nter of 2011 at both the ACMF and Osceola study sites ( Table 4 1). Overstory vegetation was sampled using a 15 m radius circular plot (0.07 ha) centered on the middle R s sample collar. The following field parameters with abbreviation and unit were record ed for each R s sample collar per plot: linear distance (m) from the sample collar to the nearest tree with a diameter at 1.3 m height (DBH) > 10 cm (Dnearest), diameter (cm) at breast height of the nearest tree to the sample collar (DBH), linear distance ( m) from the sample collar to the nearest palmetto, mean litter or masticated vegetative depth (cm)

PAGE 149

149 from three measurements within 30 cm of the sample collar (Litter), mean duff depth (cm) from three measurements taken within 30 cm of the sample collar (Duf f) and total mean duff and litter depth (cm) from three measurements taken within 30 cm of the sample collar (DL). The following stand condition parameters with abbreviation and units were recorded one time per sample plot: total basal area (m 2 ha 1 ) (BA ) a nd stand density (trees ha 1 ) (TPH) The following soil characteristics were measured one time in all plots: soil organic matter (SOM) (%) and total soil carbon (TC) (%) Three soil sub samples per plot at two soil depths from the mineral layer (0 5 cm and 5 10 cm) were collected for analysis. Soil samples were collected using a 2.22 cm AMS soil sampler after removing litter and duff layers (AMS, Inc., American Falls, Idaho, USA). To account for the spatial varia bility of soils within the plot the three sub samples per plot were bagged per depth class and homogenized. Soil samples for the Osceola and ACMF study areas were collected in the winter of 2011 approximately ten months following the burn treatment in the Osceola and approximately seventeen months following mastication in the Osceola. Samples were bagged and shipped to Waters Agricultural Laboratories, Inc. (Camilla, Georgia, USA) for analysis. Analysis The data collected from each study site were analyzed separately A t the ACMF study site prescribed fire t reatments were analyzed as random samples representing two treatment sites. At the Osceola study site, treatments were analyzed as a randomized complete block design with understory vegetative fuel treatment type (Osceola) as the main t reatment s For each month, daily measurements of R s T s and M s per soil collar were averaged, and the nine soil collar means were then averaged to

PAGE 150

150 produce a plot level mean for each treatment and month. This resulted in a sample size of three for each t reatment (Osceola total n = 12 and ACMF total n = 6) per month. One way (ACMF) and two way (Osceola) r epeated measures analysis of variance (ANOVA) was used to test for differences in monthly means of R s T s and M s among treatments at each study site. S ignificant treatment effects were identified at p value < 0.05. To assess differences between field parameters by treatment at each study site one way ANOVA tests (ACMF) were used. Where significant differences were identified in the ANOVA differences among treatment s were analyzed With treatments ignored and all monthly means pooled, linear regression were used to assess for relationships between overall study per iod mean plot R s rates and T s M s and field parameters following Gough et al. (2004). Additional linear (Equation 4 1) and nonlinear (Equation 4 2) regression models were developed per treatment and measurement season (growing vs. dormant) to assess the influence of treatments on the relationships between monthly per plot mean R s rates and T s M s and the field parameters listed in Table 4 1. At both study sites the growing season was defined as the months of March September while the dormant season wa s defined as October February (Gholz and Clark 2002). N on linear models of the relationship s between R s rates and T s and M Temp per entire study period and season were explored using an exponential equation (Equation 4 2) frequently used to describe the response of R s rates to soil temperature (Lundegardh, 1927; Samuelson et al., 2004; Concilio et al., 2005; Kobziar and Stephens, 2006) Following Samuelson et al. (2004) and Ryu et al. (2009) multiple regression using a forward step wise procedure was us ed to develop

PAGE 151

151 models per study site and treatment of monthly mean R s rates using Equation 4 3, utilizing measured parameters that best explained the observed variability in R s rates (using R 2 and p value), while minimizing multicollinearity and BIC scores. ( 4 1) or ( 4 2) ( 4 3) Coef ficients 0 1 2 i were estimated through regression analysis. Residuals of regressions were checked for normality and heteroscedasticity and where necessary model terms were transformed to meet assumptions. 1 estimates developed using Equa tion 4 2 were used to estimate the Q 10 value per treatment season, and study site using Equation 4 4 following Kobziar and Stephens (2006) (Lundegardh, 1927). The Q 10 value is often reported in studies of R s to describe the response of R s to a 10 C chan ge in soil temperature (Luo and Zhou, 2006). ( 4 4) The linear models developed per study site and treatment (using Equation 4 1) of the relationships between soil temperature (T s ) and R s rates were used to estimate hourly, monthly, and total annual soil carbon fluxes following S amuelson et al. (2004). The hourly 10 cm depth soil temperature ( C ) measurements from the Macclenny, Florida and Putnam Hall, Florida FAWN stations were used as model input. All statistical analyses were performed using JMP 9.0 (SAS Institute, Cary, NC USA).

PAGE 152

152 Results Treatment Effects Vegetative conditions varied significantly (p < 0.05) by treatment within the study area s (Table 4 2). A t the ACMF site, significantly higher basal area (27.03 m 2 ha 1 ) stand density (773.33 tree ha 1 ) duff (2.8 7 cm), and litter depth (4.11 cm) were observed in the long unburned treatment (40YR) than in the more frequently burned treatment (3YR). Analysis of soil samples at the ACMF at either the 0 5 cm depth or 5 10 cm depth found no significant differences between treatments in SOM (%) or TC (%). A t the Osceola site, there were no significant (p < 0.05) differences between treatments in stand basal area stand density, or duff depths. Prescribed burning was shown to significantly reduce litter depth in th e burn (1.57 cm) and mow+burn (1.18 cm) treatments relative to the mow (3.50 cm) and control (4.50 cm) treatments (Table 4 2). At the Osceola site like the ACMF site, no significant treatment effects were observed for SOM or TC, for either sample depth. At both study sites soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and soi l temperature ( T s ) (C) for all treatments tended to be highest during the late spring, summer, and early autumn months and lowest during the winter months (Figure 4 4 and Figure 4 5) Soil moisture content across all treatme nts and study sites was highly variable over time, with M s content generally highest during the winter and fall months and lowest during the summer months at the Osceola study site, while M s content for both treatments at the ACMF study site appeared to be strongly influenced by regional drought conditions as indicated by the Palmer Drought Severity Index (Figure 4 6). At the Osceola site over 2,800 soil CO 2 efflux rate measurements were taken during the twelve month sampling period, with plot level monthly mean R s rates ranging

PAGE 153

153 from 1.16 8.73 2 m 2 sec 1 Repeated measures ANOVA found monthly mean soil CO 2 efflux rates were not significan tly different between Osceola treatments (F = 0. 86 p = 0. 4985 ) (Table s 4 3 and 4 4) (Figure 4 7) Soil CO 2 eff lux rates did vary significantly by month (F = 35.69 p < 0.0001) but did not show an interaction between treatment effect and time (treatment x month) (F = 0.66 p = 0.9123). While not significantly different, t he lowest mean R s rates were generally in the burn treatment (3.4 4 2 m 2 sec 1 ) with some monthly variation in the order of treatments observed ( Table 4 4 ) ( Figure 4 4 ). When treatment effects on R s rates were assessed separately by season (growing vs. dormant), there were no significant differences observed, and only the effect of time (month) was significant for either season (Table s 4 3 and 4 5). At the Osceola study site, s oil temperature ( T s ) (C) ranged from 13.71 25.76 C during the entire study period and varied significantly by sample month (F = 321.78 p < 0.0001) and treatment (F = 11.42 p = 0.0 029) (Table 4 3). Mean overall soil temperature was significantly higher in all treatments relative to the control with no other significant differences between treatments observed (Table s 4 3 and 4 4) (Figure 4 7). The effect of treatment did not vary significantly with time ( treatment x month) (F = 1.30 p = 1932 ). I n the growing season all treatments had significantly higher mean T s (F = 31.86 p < 0.0001 ) relative to the control, while the mow+burn treatmen t recorded significantly warmer soil temperatures than the mow only treatment ( Table 4 3 and Table 4 5 ). In contrast, d uring the cooler dormant season treatment did not have a significant effect on soil te mperature ( Table s 4 3 and 4 5 )

PAGE 154

154 Soil moisture con tent at the Osceola study sites ranged from 0. 0 4 0. 27 m 3 /m 3 during the study period and varied significantly by sample month (F = 22 57 p < 0.0001) and treatment x time (F = 1.88 p = 0.0 106 ), but not treatment (F = 2. 33 p = 0. 1503 ) ( Table s 4 3 and 4 4 ) ( Figure 4 7) During the seasonal assessment, no significant differences in M s were found between the treatment s during either the growing season (F = 2.68 p = 0.12) or the dormant season (F = 1.24 p = 0.36) (Table 4 5) S oil moist ure content over all stu dy and monthly means showed a consistent non significant trend among treatments as mow+burn generally had the highest mean M s while the control generally had the lowest mean M s with variation observed between the burn and mow treatments. ( Table s 4 3 and 4 5 ) ( Figure 4 4). At the ACMF study site over 1860 individual soil CO 2 efflux rate measurements were taken during the 14 month sampling period with plot level monthly mean R s rates ranging from 1.30 6.34 2 m 2 sec 1 (Figure 4 5) Repeated measures ANOVA of monthly mean R s rates analyzed for the entire study period found no significant differences between treatments (F = 0. 3 5 p= 0.5 888 ) with the overall mean R s rate in the 3YR treatment (4. 12 CO 2 m 2 sec 1 ) slightly lower than the mean R s rate in the 40YR treatment (4. 25 2 m 2 sec 1 ) ( Table s 4 3 and 4 4 ) (Figure 4 8) The effects of time (sample month) on R s rates at the ACMF study site was highly significant (F = 63.08 p < 0.0001) whil e the interaction of time and treatment (treatment x time) on overall mean R s rates was also significant (F = 2.43 p = 0.0187, respectively) (Table 4 3). The season specific assessments (growing vs. dormant) found that mean R s rates during the growing sea son were high er, though not significantly in the 3YR treatments 2 m 2 sec 1 ) than the 40YR treatments (4. 37 2 m 2 sec 1 ) (Table 4

PAGE 155

155 3 and Table 4 5). In contrast, during the dormant season R s rates were significantly 2 m 2 sec 1 ) than th e 40YR treatments (4.00 2 m 2 sec 1 ) (F = 11.15 p = 0.0288 ) ( Table s 4 3 and 4 5). Plot level monthly mean s oil temperature s at the ACMF ranged from 12.49 26.59 C during the study period Repeated measures ANOVA found that T s differed significant ly by sample month (F = 900.65 p <0.0001) and treatment x time interaction (month x treatment) (F = 31 09 p < 0.0001) but not by treatment type (F = 5.74 p = 0. 0746 ) (Table s 4 3 and 4 4) (Figure s 4 5 and 4 8). A similar analysis found that during the gro wing season mean soil temperature in the 3YR treatment (20.81 C ) was significant ly (F = 11. 23 p = 0.0286 ) warmer than in the 40YR treatment (19.76 C ) (Table s 4 3 and 4 5). There were no significant differences in T s due to treatment during the cooler do rmant season (F = 0.00 p = 0.9 789 ) ( Table 4 3 ). Overall plot level monthly mean soil moisture conte nt at the ACMF ranged from 0. 02 0. 38 m 3 /m 3 during the study period A repeated measures ANOVA found that M s differed significantly by sample month (F = 45 57 p < 0.0001) treatment x time interaction (F = 3 80 p = 0.000 8 ) and treatment (F = 7 88 p = 0.0 484 ) with overall study period mean M s highest in the 3YR treatment (0.11 m 3 /m 3 ) and lowest in the 40YR treatment (0.07 m 3 /m 3 ) (Table 4 3 and Table 4 4) (F igure s 4 5 and 4 8). When M s was assessed by season, significant differences (F = 8.64 p = 0.0434) between treatments were identified only in the growing season with the highest mean M s in the 3YR treatment (0.14 m 3 /m 3 ) and the lowest in the 40YR treatme nt (0.08 m 3 /m 3 ) (Table s 4 3 and 4 5). In general, the plot of the monthly means indicated that M s tended to be higher in the 3YR treatment than in the 40YR treatment, with the difference between

PAGE 156

156 treatments most pronounced during the first four months of s ampling prior to the establishment of a prolonged regional drought ( Figure s 4 5 and 4 6 ). Overall Drivers of Soil CO 2 Efflux When treatments were ignored at each study site and all monthly mean plot values oefficients and linear regressions were used to identify broad overall relationships between R s T s and M s and plot vegetative and indicated positive relationships b etween R s and T s (0.63) and R s and M Temp (0.52) (Table 4 s and M Temp were not surprisingly strongly correlated (0.77) at the Osceola study site (Table 4 6). Also surprisingly, R s at the Osceol a study site indicated a negative relationship with soil moisture content ( 0.13) while also demonstrating a positive relationship with monthly precipitation patterns (0.35). Vegetative conditions were shown to have a small influence on overall R s rates a s litter depth (0.06), duff depth (0.08), stand density (0.00), and basal area (0.11) were only weakly correlated with R s rates at the Osceola study site. Negative correlations observed between R s rates and distance to nearest tree ( 0.06) and distance to nearest palmetto ( 0.18) suggested that there was a small yet positive influence of the proximity of measurement points to trees and palmettos. In the Osceola study site, linear regressions of pooled monthly mean values, soil temperature (R 2 = 0.40 p < 0 .0001) and M Temp (R 2 = 0.27 p < 0.0001) were also positively linearly correlated with overall mean R s rates, while plot level vegetative characteristics such as basal area (R 2 = 0.01 p = 0.1772), stand density (R 2 = 0.00 p = 0.9991), distance to nearest t ree (R 2 = 0.00 p = 0.4682), and distance to nearest palmetto (R 2 = 0.03 p = 0.0274) were not correlated with R s rates (Figure 4 9). At the

PAGE 157

157 relationship between R s and T s (0.89 ) and R s and M Temp (0.82). Similar to the Osceola study site, a negative relationship was identified between R s and M s ( 0.25) while a contrasting positive relationship was identified between R s and monthly n coefficients indicated that overall mean R s rates were not correlated with vegetative characteristics including stand density (0.01), basal area ( 0.03), distance to nearest palmetto (0.00), duff depth (0.02), and litter depth (0.05) (Table 4 7). Like R s T s means were weakly correlated with vegetative and characteristics, while M s means were negatively associated with T s ( 0.50), stand density ( 0.25), basal area ( 0.23), duff depth ( 0.28), and litter depth ( 0.38) and showed a positive association wit h the distance to the nearest tree (0.27) and monthly total precipitation (0.40). In the ACMF study site linear regressions of pooled monthly mean values soil temperature (R 2 = 0.80 p < 0.0001) and monthly mean air temperature (R 2 = 0.68 p < 0.0001) were positively linearly correlated with R s rates while M s (R 2 = 0.06 p = 0.0332) and monthly precipitation (R 2 = 0.06 p = 0.0414) were significantly, but only weakly correlated with R s rates (Figure 4 11). Similar to the results of the Osceola study site simp le linear regressions of the pooled ACMF study site data found no significant correlations between R s rates and basal area, stand density, distance to nearest tree, distance to nearest palmetto, or duff and litter depth (Figure s 4 11 and 4 12). Treatment Specific Drivers of Soil CO 2 Efflux To assess the influence of T s M s and vegetative and meteorological conditions on monthly mean R s rates within treatments at each study site, simple linear regression models (Equation 4 1) wer e developed for each parameter and treatment (Table 4 8).

PAGE 158

158 At the ACMF study site, simple linear regression models identified significant positive linear relationships between R s rates and soil temperature ( T s ) in the 3YR (R 2 = 0. 83 p < 0.0001 ) and 40YR tr eatments (R 2 = 0. 85 p < 0.0001 ) (Table 4 8) (Figure 4 14). Similarly, the relationships between R s rates and monthly mean air temperature were also significantly linearly related for both the 3YR and 40YR treatments ( R 2 = 0. 74 p < 0.0001 and R 2 = 0. 62 p < 0.0001, respectively ) (Table 4 8) (Figure 4 14). Soil moisture content was significantly negatively linearly correlated with R s in the 40YR treatment (R 2 = 0. 12 p = 0.0380 ) but not in the 3YR treatment (R 2 = 0. 04 p = 0.2565). In the ACMF simple linear regression models no other plot level vegetative or meteorological characteristics were significantly linearly correlated (p < 0.05) with R s or had R 2 > 0.10. At the Osceola study site similar to the results of the ACMF study site the simple linear regr ession models by treatment identified significant positive relationships between soil CO 2 efflux rates ( R s ) and soil temperature ( T s ) (R 2 = 0. 33 0.58) and monthly mean ambient air temperature (Temp) (R 2 = 0. 24 0.34) ( Table 4 8 ) (Figure 4 13). Only a w eak linear relationship was identified between R s rates and soil moisture content (M s ) (R 2 = 0. 02 0.12) at the Osceola treatments, with one treatment having a positive relationship with M s (Control) and the remainder having a negative relationship with M s (Burn, Mow, Mow+Burn) (Table 4 8). Some of the vegetative characteristics were significantly linearly correlated with monthly mean R s rates in the Burn treatment (R 2 = 0.17 0.20), while all other vegetative characteristics and treatments had non signi ficant (p > 0.05) or low correlation coefficient (R 2 < 0.10) relationships with R s or (Table 4 8).

PAGE 159

159 In addition to the simple linear models, n onlinear exponential models (Equation 4 2) were used to further explore the relationship s between monthly mean R s a nd T s by study site and treatment. At the ACMF study site, nonlinear regression indicated a strong positive relationship between R s and T s for both the 3YR (R 2 = 0.80) and 40YR treatment (R 2 = 0.82) (Table 4 9) (Figure 4 16). At the Osceola study site th e fit of the R s and T s nonlinear regression models (R 2 = 0.32 0.56) were similar to that of the simple linear regression models (R 2 = 0.32 0.56) described previously (Table 4 9) (Figure 4 15). Nonlinear m 0 1 ( Equation 4 2 ) were similar to estimates reported by Samuelson et al. (2004) and Kobziar and Stephens (2006) (Table 4 9). Seasonal Drivers of Soil CO 2 Efflux To assess seasonal variations in the relationship s between R s and T s M s and plot vegetative and meteorological characteristics, additional simple linear (Equation 4 1) and nonlinear (Equation 4 2) regression models were developed per study site, treatment, and season ( growing and dormant) (Table s 4 10 and 4 11 ) (Figure s 4 17, 4 18, 4 19, and 4 20). At the ACMF study site, models for both treatments indicated that the relationships between R s and T s varied seasonally, with both linear (Table 4 10) (Figure 4 19) and non linear models (Table 4 11) (Figure 4 20) showing that soil temperature explained mu ch more of the variability in R s rates during the growing season (R 2 = 0.89 0.90) than during the dormant season (R 2 = 0.50 0.53). In contrast, linear models for both treatments identified negative relationships between R s rates and M s with M s explai ning much more of the variability in R s rates during the dormant season (R 2 = 0.74 0.50) than during the growing season (R 2 = 0.29 0.19) (Table 4 10).

PAGE 160

160 At the Osceola study site, linear and nonlinear models of all treatments found that the relationships between R s and T s and R s and M s varied seasonally (Table 4 10 and Table 4 11). At the Osceola study site, in contrast to the ACMF study site, positive linear models of T s explained much more of the variability in R s rates during the cooler dormant season (R 2 = 0.69 0.79) than during the warmer growing season (R 2 = 0.30 0.60) (Table 4 10) (Figure 4 17). Similarly, positive non linear models of T s explained much more of the variability in R s rates during the dormant season (R 2 = 0.64 0.76) than durin g the growing season (R 2 = 0.30 0.58) (Table 4 11) (Figure 4 18). For either season or modeling type (linear or non linear), the burn treatment recorded the weakest relationship between R s and T s while the mow treatment recorded the strongest relations hip (Table 4 10 and Table 4 11). In the soil moisture content linear models, monthly mean R s was more closely correlated (negatively) with M s during the dormant season (R 2 = 0. 24 0. 67 ) than during the growing season ( R 2 = 0. 09 0. 53 ) in all treatments except for the burn treatment (Table 4 10). In the burn treatment, R s was more closely correlated with M s during the growing season (R 2 = 0.44) than the dormant season (R 2 = 0.11) (Table 4 10). Multiple Regression Models Multiple linear r egression models using Equation 4 3 were developed per study site, treatment, and season to identify the influence of treatment (Table 4 12) and season (Table 4 13) on the drivers (Table 4 1) of soil CO 2 efflux rates. Models were developed using a forward step wise approach (minimum parameter input and retention p < 0.05) using the stand and plot characteristics described in Table 4 1 as potential parameters. Models were developed to minimize BIC score and parameter multicollinearity while maximizing mode l coefficient of variation (R 2 ) The ACMF study

PAGE 161

161 site all season multiple linear regression models explained a significant amount of the observed variation in R s rates for both the 3YR (R 2 = 0.90 p < 0.0001) and 40YR treatments (R 2 = 0.89 p < 0.0001) (Tabl e 4 12). R s models for both treatments identified T s and M s as significant terms while the 3YR model identified a significant negative relationship with distance (m) to the nearest palmetto. The ACMF season specific multiple linear regression models were slightly better at predicting R s rates during the growing season (R 2 = 0.93 0.94) than during the dormant season (R 2 = 0.81 0.83), although all models were significant (p < 0.05) (Table 4 13). Surprisingly, neither the 3YR nor the 40YR dormant season model of R s identified T s as a significant parameter, however both treatments identified significant negative relationships with precipitation (Table 4 13). The growing season ACMF multiple linear regression models identified significant relationships wi th T s and soil organic matter (%) (5 10 cm depth) in the 3YR treatment and T s and M s in the 40YR treatment (Table 4 13). The amount of variation in R s rates explained by the Osceola study site all season multiple linear regression models varied by treatm ent and ranged from (R 2 = 0.36 0.72) (Table 4 12). Similar to the linear and non linear models, the mow+burn treatment model explained the least amount of the variation in R s (R 2 = 0.36 p = 0.0006) while the control treatment model explained the greates t amount of the observed variation in R s rates (R 2 = 0.72 p < 0.0001) (Table 4 12). Soil temperature (T s ) explained the majority of R s rate variability in each of the Osceola all season models with additional significant model terms (M s stand tree densit y, and stand basal area) significant in different treatment s (Table 4 12). The Osceola season specific models explained more of the variation in R s rates during the cooler dormant season (R 2 = 0.85 0.93) than during the

PAGE 162

162 warmer growing season (R 2 = 0.53 0.85) (Table 4 13). During both the growing season and the dormant season the mow+burn treatment models explained the least amount of observed variation in R s (R 2 = 0.53 and R 2 = 0.85, respectively) while the control model explained the most in the dor mant season (R 2 = 0.93) and the burn treatment model explained the most during the growing season (R 2 = 0.84) (Table 4 13). In the Osceola season specific multiple linear regression procedures model terms identified as significant during the forward step wise process differed by treatment type and season. During the growing season T s was significant only in the control and m ow models, while the other treatment models included M s monthly mean air temperature, precipitation, and distance to nearest palmett o (Table 4 13). During the dormant season T s explained the majority of the variation observed in R s rates in all models with subsequently added parameters explaining much less of the variation in R s The model parameters other than T s identified as signi ficant (stand tree density, stand basal area, soil moisture content, and distance to nearest tree) provided evidence for the roles of additional drivers of R s beyond T s and M s ; although overall patterns in the relationships were not clear (Table 4 13). Tem perature Response For each study site, treatment, and season, the model estimates from Equation 4 2 were used to estimate Q 10 values using the Q 10 model (Equation 4 4) (Lundegardh, 1927) (Table s 4 9 and 4 11). Q 10 values are used to describe the incremental response of R s to a 10 C change in soil temperature (Lundega rdh, 1927; Kobziar and Stephens, 2006). At the ACMF study site, in the all seasons models, Q 10 values for the 3YR treatment (Q 10 = 2.14) were lower than the Q 10 values in the 40YR treatments the (Q 10 = 2.85) (Table 4 9). At the Osceola study site, estima ted Q 10 values in the all seasons

PAGE 163

163 models were highest in the control (Q 10 = 2.14) and lowest in the mow+burn treatment (Q 10 = 1.65) (Table 4 9). The Q 10 values in the season specific models at the ACMF study site and Osceola study site ranged from Q 10 =1. 62 2.90 and Q 10 = 1.63 2.51 respectively (Table 4 11). At the ACMF site, Q 10 values tended to be slightly lower in the dormant season than the growing season, while at the Osceola study site, the opposite was observed, as Q 10 values in all treatment t ypes and control were higher during the dormant season than the growing season. Estimated C arbon F lux Following Samuelson et al. (2004) total soil c arbon (C) flux was estimated per hour for every 24 hour period and summed to compare total monthl y and annual soil C flux es per treatment at each study site. Soil c arbon fluxes were estimated using treatment specific linear models (Table 4 8) of the relationship between soil CO 2 efflux CO 2 m 2 sec 1 ) and soil temperature (Table 4 11) Ac tual h ourly 10 cm depth soil temperature measurements ( C) recorded at the Florida Automated Weather Network (FAWN) station s in Macclenny, Florida near the Osceola study site, and Putnam Hall, Florida near the ACMF study site were used as model input s to p redict flux rates Model input soil temperatures for the Austin Cary Forest were recorded March 2010 February 2011 while input soil temperatures for the Osceola study site were recorded February 2011 January 2012. Predicted hourly soil CO 2 fluxes ( mol CO 2 m 2 sec 1 ) were then converted to hourly soil C fluxes (g C m 2 hr 1 ) and summed to estimate monthly and annual carbon flux es per treatment and study site At the Osceola study site total estimated monthly soil carbon flux (g C m 2 month 1 ) varied monthly and seasonally. During the growing season, soil carbon flux was

PAGE 164

164 greatest in the control sites and lowest in the burn only treatments and mow+burn treatments (Figure 4 21). During the cooler dormant season months of December, January, and Februar y, monthly total soil carbon flux in the control treatments reduced to lower than the other three treatment types (Figure 4 21 ). When soil carbon fluxes were totaled for the entire year, estimated annual soil C flux was greatest in the control units (1.89 kg C m 2 yr 1 ), and lowest in the mow+burn treatment (1.64 kg C m 2 yr 1 ), with the burn (1.69 kg C m 2 yr 1 ) and mow treatments falling in between (1.67 kg C m 2 yr 1 ) (Figure 4 2 2 ). At the ACMF study site total estimated soil carbon flux varied both m onthly and seasonally (Figure 4 23). During the growing season the estimated soil carbon flux was highest during the months of July and August with the greatest estimated C flux in the 40YR treatment (218.62 g C m 2 month 1 and 219.54 g C m 2 month 1 resp ectively) and the lowest C flux in the 3YR treatment (184.63 g C m 2 month 1 and 185.28 g C m 2 month 1 respectively). During the dormant season when overall soil carbon fluxes were at their lowest rates, the trend reversed, and the 3YR treatment had the greatest C fluxes (Figure 4 23). When estimated soil carbon fluxes for the entire year were summed, the 40YR prescribed fire interval resulted in a 7% higher soil carbon flux (1.61 kg C m 2 yr 1 ) than the 3YR treatment (1.51 kg C m 2 yr 1 ) (Figure 4 24). Discussion The range of monthly mean soil CO 2 efflux rates ( mol CO 2 m 2 sec 1 ) recorded at the Osceola (1 .16 8.73 mol CO 2 m 2 sec 1 ) and ACMF (1.30 6 34 mol CO 2 m 2 sec 1 ) study sites were similar but higher than those reported in many other publish ed studies of R s rates. In a similar study of mechanical fuels mastication treatments, prescribed fire, and mechanical treatments followed by fire in a mixed conifer plantation

PAGE 165

165 in California, USA, Kobziar and Stephens (2006) reported growing season R s rat es ranging from 2.37 4.55 mol CO 2 m 2 sec 1 In another California study, Tang et al. (2005) also reported similar mean R s rates (3.26 3.78 mol CO 2 m 2 sec 1 ) for thinned and un thinned ponderosa pine plantations. The range s of R s rates reported in the Osceola and ACMF study sites were also similar to those reported by Fang et al. (1998) and Ewel et al. (1987a) for mature slash pine plantations in north central Florida. The higher R s rates recorded in our studies relative to those mentioned previously from the western USA, w ere likely related to the relatively high mean annual temperatures, frequent precipitation, and long growing seasons in our sites. Effects of Prescribed Fire and Mechanical Fuels Mastication At the Osceola and ACMF study site neither mechanical fuels mastication, prescribed burning, nor mechanical fuels mastication followed by prescribed burning significantly altered overall mean soil CO 2 efflux rates relative to control (Osceola) or prolonged fire exclusion (AC MF). These results are similar to Kobziar (2007) who found that a single mechanical fuel mastication treatment had no significant effect on soil CO 2 efflux rates in a Sierra Nevada mixed conifer plantation in California, USA. In a similar western US stud y, Concilio et al. (2005) found no significant effect of prescribed burning on soil CO 2 efflux rates in the mixed conifer Teakettle Experimental Forest. In a related study of the effects of forest thinning on soil CO 2 efflux rates and soil conditions, Tan g et al. (2005) also found no significant changes in overall mean soil CO 2 efflux rates following treatment. The lack of a treatment effect on R s rates at either of our study sites was surprising given the multiple effects that prescribed fire and mechani cal fuels treatments can have on the autotrophic (R a ) and heterotrophic (R h ) sources of soil CO 2 efflux. Both mechanical fuel treatments and prescribed fire kill, consume, or damage

PAGE 166

166 live understory vegetation thereby reducing sources of root respiration ( R a ). At the same time, such treatments provide heterotrophic soil microorganisms with fresh labile carbon in the form of dead plant roots, potentially increasing R h contributions to R s Surprisingly mechanical fuels mastication at the Osceola study site did not result in increased duff and litter depth compared to the control, burn, and mow+burn treatments (Table 4 2). Contrasting with those results were qualitative observations from the mowed sites that reported a distinct intact litter layer comprised of fractured vegetative material covering much of the forest floor (Figure 4 2). These results contrast with those of Kobziar (2007), who found increases in litter and duff depth following mechanical fuels mastication in a California, USA mixed conifer pi ne plantation. Recent research by Kreye (2012) found that because the masticated material in pine flatwoods is predominantly foliar, rather than woody (as is the case in many of the western USA studies), its decay rate is much faster, its packing ratio hi gher, and its overall contribution to litter depth only temporarily significant. At the ACMF study site, prolonged fire exclusion in the 40YR treatments resulted in significantly increased duff and litter depths compared to the frequently burned (3YR) trea tments (Table 4 2). Similar to the ACMF 40YR treatments, Varner et al (2005) found that prolonged fire exclusion in southern pinelands resulted in deep duff and litter layer development while frequently burned sites similar to the 3YR treatments had littl e organic matter accumulation on the surface. Others have suggested that masticated plant material deposited on the forest floor and surface litter represents a potential source of labile carbon for heterotrophic soil microorganism metabolism and respirat ion (R h ) (Kobziar and Stephens, 2006; Ryu et al., 2009) Supporting this hypothesis are

PAGE 167

167 results from studies where litter and duff depths have been positively associated with soil CO 2 efflux rates (Concilio et al., 2005; Kobziar and Stephens, 2006) In a ddition experimental manipulations in multiple studies have found soil CO 2 efflux rates to increase significantly following litter additions (Bowden et al., 1993; Chemidlin Prvost Bour et al., 2010; Sulzman et al., 2012) In addition, a partitioning stu dy of soil CO 2 efflux sources within a 29 year old slash pine plantation in North Florida, Ewel et al. (1987b) reported that 48% of measured R s could be attributed to roots and microbial decomposition in the litter and humus horizons of the forest floor. In contrast, our results found little support for the association between litter depth and soil CO 2 efflux rates however this may have been the result of our data analysis dealing with plot level means instead of individual sample point means or our inabil ity to differentiate between sources of R s Further study utilizing 13 C isotopic sampling may facilitate better understanding of the connectivity between natural surface litter layers accumulated over long fire free periods and mechanically derived surfac e litter layers following mastication treatments and the sources of soil CO 2 efflux rates. The lack of a detected R s response to the treatments could be due to the inability of the monthly sampling protocol employed in this study to capture short term trea tment responses. This is supported by a recent dissertation study of the Florida Everglades by Medvedeff (2012) that found a soil microbial response (in the form of altered R h ) was detectible only within two days of prescribed burning. The Medvedeff (201 2) study reported that subsequent post burn sampling of R h in burned and unburned sites weeks and months following treatment revealed no significant differences between treatments. Other studies including a 13 C isotope tracing experiment in a coniferous f orest in

PAGE 168

168 northern Sweden, Ekblad and Hogberg (2001) and a 13 C tracing experiment in a 15 y ear old loblolly pine plantation in North Carolina, USA (Andrews et al. 1999) have reported soil CO 2 efflux responses to treatments in as little as 1 4 days. We sugg est, similar to Medvedeff (2012), that future R s sampling protocols employ more temporally intensive measurements in the period immediately following treatment while maintaining monthly long term sampling to capture annual and seasonal variability. Quanti fying short lived responses may seem insignificant at the ecosystem level, but over broad spatial scales even transient CO 2 fluxes may be important for future landscape level carbon budgeting and modeling. It is also possible that compensatory responses fr om R a and R h sources following treatment in our studies masked any detectible overall treatments effects on R s rates. Such a situation could occur if reduced understory vegetation in the frequently burned ACMF treatment and the burned and mowed treatments at the Osceola study site led to reduced R a contributions to R s while simultaneous increases in soil microbial consumption of the recently killed roots led to similar elevated R h contributions to R s Given that previous studies partitioning root and m icrobial sources of CO 2 efflux in slash pine plantations have shown the contributions of the two sources to be similar (Ewel et al., 1987b) this decrease and subsequent increase in the sources of R s could offset statistically detectible changes in R s Oth ers who have investigated the effects of forest management practices on soil carbon dynamics have proposed similar explanations following studies where little or no treatment effects on R s rates were observed (Shan et al., 2002; Tang et al., 2005; Ryu et a l., 2009).

PAGE 169

169 Our results from the Osceola study site found that while prescribed burning and mechanical mastication treatments did not have a significant effect on soil CO 2 efflux rates, they did result in increased soil temperature relative to control sites Similarly at the ACMF study site, while prolonged fire suppression in the 40YR treatment was shown not to affect mean total R s rates, it was shown to reduce soil moisture content relative to sites (3YR) managed with frequent prescribed fire. Previous s tudies of forest management practices have documented similar effects on soil conditions. In a study of a mixed conifer forest in California, USA, Ryu et al. (2009) reported that prescribed burning and forest thinning treatments increased soil temperature and moisture content while simultaneously reducing soil CO 2 efflux rates. In another example, following a heavy thinning treatment in a ponderosa pine plantation in California, USA, Tang et al. (2005) found that thinning increased forest soil temperature and soil moisture while having no clear overall effect on mean R s rates. The elevated soil temperatures observed in all treated sites (burn, mow, mow+burn) relative to the control sites at the Osceola study site were likely due to the effect of prescribe d fire and mechanical fuels mastication treatments on canopy cover and forest floor radiation exposure. Other studies have shown that forest cover (Michelsen Correa and Scull, 2005), forest management practices (Castro et al. 2000), and fire (Neary, 1999 ; Medvedeff, 2012) can significantly influence long term soil temperatures among sites due to changes in canopy and vegetative cover. At the ACMF study site, the elevated soil moisture content observed in the frequently burned treatments (3YR) relative to the long unburned treatments (40YR) may have been due to changes in vegetative abundance and composition and the corresponding demands of such vegetation for soil water. At

PAGE 170

170 the ACMF study site, stand basal area (m 2 ha 1 ), density (trees ha 1 ), duff depth (cm), and litter depth (cm) were all significantly greater in the 40YR treatment than in the 3YR treatment (Table 4 2). In addition, though not quantified, observations from the ACMF sites suggested that palmetto height and density was much greater in th e long unburned sites than in the frequently burned sites (Figure 4 3). These results are similar to those reported by Burger and Pritchett (1988) and Castro et al. (2000) following studies of the effects of silvicultural treatments that reduced site vege tation on soil moisture content in North Florida slash pine plantations. Soil CO 2 Efflux Response to Temperature Fluctuations Soil temperature was generally the strongest assessed driver of soil CO 2 efflux rates in o ur study at both study sites. When treatment types were ignored and all data per study site pooled of all parameters, T s had the highest correlation with R s at both the Osceola (0.63) and ACMF (0.89) study si tes (Table s 4 6 and 4 7). In the treatment specific all season linear regression models, T s explained more variability in R s at the ACMF study site (R 2 = 0.83 0.85) than at the Osceola study site (R 2 = 0.33 0.58) (Table 4 8). Multiple regression mode ls by treatment type at both study sites also identified soil temperature as explaining much of the variability in R s rates (Table s 4 12 and 4 12). These results are similar to three previously published studies of slash pine plantations in north central Florida, Clark et al. (2004) (R 2 = 0.49 0.78), Ewel et al. (1987a) (R 2 = 0.75), and Fang et al. (1998) (R 2 = 0.96) that all reported similar correlations between soil temperature and R s rates. At the Osceola study site, the burn and mow+burn treatments tended to reduce the amount of variability in R s rates explained by T s in linear and non linear models in comparison to the control and mow treatments (Table s 4 8 and 4 9). This

PAGE 171

171 may have been due to reductions in the relative importance of temperature in governing R s production as plants and soil microorganisms responded to changes in microsite conditions and nutrient availability following fire (Medvedeff, 2012). Further study may show whether additional time since fire in the burned treatments leads to an increase in the amount of R s variability explained by T s Positive correlations between soil CO 2 efflux rates and soil temperature have raised concerns regarding soil carbon fluxes under elevated temperatures due to global climate change (Rustad et al., 2000; Schlesinger and Andrews, 2000) Other factors associated with global climate change including elevated atmospheric CO 2 concentrations have also been associated with changes in soil CO 2 flux rates (Schlesinger and Andrews, 2000) Butnor et al. (200 3) reported increased R s rates in loblolly pine stands in North Carolina, USA during a free air CO 2 enrichment (FACE) study, suggesting that increased atmospheric CO 2 concentrations may drive positive feed back cycles leading to soil carbon loss. In a si milar study, Carney et al. (2007) reported increased belowground heterotrophic microorganism activity following CO 2 enrichment that led to increased R s rates and subsequent soil carbon losses. Our results and these suggest that research combining forest m anagement practices with experimental in situ CO 2 enrichment techniques may be exceptionally beneficial for predicting the effects of management practices on soil carbon fluxes and pools. The estimated all season Q 10 values of the soil CO 2 efflux response to changes in soil temperature across all of our study sites ranged from Q 10 = 2.14 2.85 at the ACMF study site and Q 10 = 1.65 2.14 at the Osceola study site (Table 4 9) (Lundegardh, 1927) These estimated values were similar to those reported by Fang et al. (1998) and

PAGE 172

172 Clark et al. (2004) following studies of soil carbon dynamics in north Florida slash pine plantations. Q 10 values were also similar to root specific respiration measurements taken in a north Florida slash pine plantation by Cropper and Gholz (1991 ) The seasonal and overall estimated Q 10 values from our study imply that the sources of soil CO 2 efflux rates may have varied under the different treatment types. Previous research has suggested that heterotrophic soil microorganisms are lik ely to be more sensitive to changes in soil temperature than autotrophic sources of R s (Bhupinderpal Singh et al., 2003) As Q 10 values are a measure of the temperature sensitively of soil CO 2 efflux rates, treatments with higher estimated Q 10 values woul d therefore be more likely to be driven by heterotrophic sources of R s than autotrophic sources. At the ACMF study site, Q 10 values for both the all seasons models and season specific models were highest in the 40YR treatment and lowest in the 3YR treatme nt, suggesting that the frequent fire regime of the 3YR treatment resulted in reduced heterotrophic contributions to total R s as compared to prolonged fire exclusion. Supporting this are the results of previous research by Ewel et al. (1987b) from a long unburned slash pine plantation (similar to our 40YR treatment) that reported that heterotrophic contributions to total R s (from the decomposition of forest floor litter) increased with stand age. Similar to the ACMF study site, at the Osceola study site, estimated Q 10 values in the all seasons models and in some of the season specific models were highest in the long unburned control treatment and lowest in the burn and mow+burn treatment. These results suggest that all treatments at the Osceola may have r educed R h sources of R s relative to the control, with the burn and mow+burn treatments having had the

PAGE 173

173 greatest reduction in R h These results are similar to a study of a mixed conifer forest in California, USA, where Ryu et al. (2009) suggested that R s ra tes in sites treated with prescribed fire were largely controlled by R a sources as R h sources had been reduced due to fire. The authors further suggested that R s production in sites treated with mechanical forest thinning was largely controlled by R h sour ces as R a sources had been reduced due to thinning (Ryu et al., 2009). If the trends in the relative contributions of R h to R s suggested by the estimated Q 10 values are an accurate description of the partitioning of R s in our study areas, then our results suggest that either frequent prescribed fire (ACMF 3YR), or a single prescribed fire (Osceola burn), or a single prescribed fire following mechanical fuels mastication (Osceola mow+burn) can alter soil environmental conditions sufficient to reduce the con tribution of heterotrophic soil microorganisms. However, utilizing Q 10 values as a partitioning method is not a well established practice, and the results of Bhupinderpal Singh et al., ( 2003) have been contradicted by other studies (Boone et al., 1998; Sa iz et al., 2006) In some ways our results also question the strength of the Q 10 method for partitioning R s sources. For example, in both of our study sites in the treatments in which the Q 10 values suggested that the R h sources dominated R s vegetative quantitative (Table 4 2) and qualitative (Figure 4 2 and Figure 4 3) observations also suggested that R a sources might instead dominate total R s These results emphasize the importance of future non invasive partitioning studies of the drivers of R s (Bagg s, 2006). At both the ACMF and Osceola study sites, seasonal differences in Q 10 values and the strength of the T s and R s relationship showed a distinct pattern. In all treatments at

PAGE 174

174 the Osceola study site during the dormant season Q 10 values and the fit o f R s linear and non linear models of the R s T s relationships were higher than during the growing season. We suggest that the seasonal changes in the Q 10 values and R s models were indicative of treatment independent phenological shifts in the relative co ntributions of R a and R h to R s Previous research from partitioning studies has shown that during periods of aboveground vegetative growth R a contributions to R s can increase as plants allocate recent C photosynthate belowground, driving higher root main tenance, root growth, and mycorrihizal fungal respiration rates (Subke et al., 2006; Kuzyakov and Gavrichkova, 2010) In addition, previous research has also shown that during growing seasons, the T s and R s relationship can weaken as other variables such as soil moisture and available photosynthetically active radiation become more important in governing belowground C allocation by plants (Ekblad and Hogberg, 2001; Davidson et al., 2006; Wertin and Teskey, 2008) Curiously, at the ACMF study site the sea sonal trend in Q 10 values and the R s temperature relationship were opposite that of the Osceola study site. More specifically, at the ACMF study site Q 10 values and the fit of R s T s models were higher during the growing season than during the dormant season. It is not clear what could have led to this discrepancy in results between the two study sites and between the ACMF study site and existing literature. The influence of the building regional drought during both study periods may have played a rol e in altering the seasonal relationships between the biotic and abiotic drivers of R s as drought conditions were variable throughout both the ACMF and Osceola study periods. Both the results from the ACMF and the Osceola study sites support suggestions by others (Lee et al., 2003) that future models of soil

PAGE 175

175 CO 2 efflux rates and soil carbon flux should account for site and treatment specific seasonal relationships between R s and soil temperature. Soil CO 2 Efflux Response to Soil Moisture and Precipitation Other evidence from our study hinted at the possible influence of drought, precipitation, and soil moisture content on R s rates. At both t he Osceola and ACMF study sites when treatments were ignored and all plot m onthly means were pooled, R s and M s 0.13 and 0.25, respectively), while at both study sites R s and monthly precipitation were positively correlated (0.35 and 0.24, respectively (Table 4 6 and Table 4 7 ). In addition, at both study sites, the amount of site vegetation (stand density, distance to nearest palmetto, and litter and duff depth) appeared to be negatively associated with M s and positively, though very weakly associated with R s (Table 4 6 and Table 4 7). These results suggest that the inverse relationships observed between R s rates and M s may not be causal, but rather indicative of the relationships between plants, soil moisture, and the sources of R s In both of our stud y areas, these results sug gest that vegetative abundance within sample sites had a stronger effect on soil moisture content than on soil CO 2 efflux rates, as increased vegetative abundance was shown to be more strongly associated with decreased soil moisture content than increased soil CO 2 efflux rates. S ite vegetation has been shown to strongly reduce soil moisture content in in many ecosystems including flatwoods forests due to the water demand of forest vegetation (Burger and Pritchett, 1988; Castro et al. 2000). Similarly, ot her studies have also shown that the amount of aboveground vegetation has been positively associated with soil CO 2 efflux rates, as belowground carbon allocation by plant roots leads to increases in R a and aboveground litter and organic matter leads to inc reases in R h (Luan et al. 2011). The

PAGE 176

176 positive correlation between monthly precipitation and R s and M s (0.24 and 0.40, respectively) at the ACMF and R s and M s (0.35) at the Osceola, suggests that in general, both soil moisture content and the R a and R h so urces of soil CO 2 efflux likely responded positively to increased precipitation at both study sites Our observed relationships between R s M s and monthly precipitation may explain why given that our study occurred during an extended regional drought, so me aspects of our results conflict with previous studies. Effects of Treatment on Soil Carbon Flux Our monthly soil carbon flux (g C m 2 month 1 ) estimates showed little treatment induced monthly and seasonal variations in soil ca rbon emissions (Figure s 4 21 and 4 23). In our study both prescribed fire and mechanical fuels mastication treatments at the Osceola study site resulted in slightly reduced estimated total annual soil carbon fluxes (Figure 2 22). In addition, at the ACM F study site the frequently burned treatment was shown to slightly reduce estimated annual soil carbon fluxes relative to fire excluded treatment (Figure 2 24) R esults from both sites are similar to estimated soil carbon fluxes ( 0.952 1 .162 k g C m 2 yr 1) following several years of continuous measurement s reported by Clark et al. (2004) along a slash pine plantation chronosequence in north central Florida. Our results were also similar to the total estimated soil carbon flux reported for a 220 day meas urement period in a 17 year old loblolly pine stand (1 047 k g C m 2 ) in North Carolina, USA (Butnor et al., 2003) As these results are only annual soil carbon flux calcul ations, the effects of prescribed fire and mechanical fuel mastication treatments on long term soil carbon fluxes and possibly more importantly, total ecosystem carbon fluxes, will likely of great interest to resource managers and researchers.

PAGE 177

177 Conclusion This study found that neither prescribed fire nor mechanical fuels mastication treatm ents significantly affected overall mean soil CO 2 efflux rates in mature flatwoods forests. Measured soil CO 2 efflux rates in all flatwoods sites varied seasonally and were largely correlated with soil temperature. Although this was not a specific partit ioning study, our estimated Q 10 values support the findings of others who suggest that the contributions of heterotrophic and autotrophic sources of soil CO 2 efflux vary seasonally (Lee et al., 2003; Xu et al., 2011) Our results suggest that further stud y of the effect of prescribed fire and mechanical fuels mastication treatments that include intensive short term and long term pre and post treatment 13 C isotopic sampling may provide further information regarding the interactions between treatments and t he sources of R s Future attempts to model soil carbon dynamics in these systems should account for the effects anthropogenic forest management activities on soil abiotic conditions and seasonal variations in the response of soil CO 2 efflux biotic and abi otic drivers. Prescribed fire, mechanical fuels mastication, and mechanical fuels mastication followed by prescribed fire were found to significantly increase mean annual soil temperature within the Osceola National Forest flatwoods sites. Future research is needed to understand whether the changes in soil temperature will ultimately lead to altered decomposition rates and soil carbon fluxes. Our results however, found no evidence of elevated soil CO 2 fluxes within one year of mastication treatment. A man agement regime of frequent 3 year dormant season prescribed fire at the ACMF flatwoods study site was shown to increase monthly mean soil moisture content relative to a management regime of long term fire suppression. Understanding the impacts of forest m anagement practices on soil moisture content may be increasingly

PAGE 178

178 important in the future given the likelihood of prolonged droughts within many parts of the southeastern US due to the effects of global climate change (Karl et al., 2009; Liu et al., 2010).

PAGE 179

179 Table 4 1. Parameters accessed for influence on soil CO 2 efflux rates at the Austin Cary Forest and Osceola National Forest, Florida, USA Parameter category Plot variable Abbreviation Measured Measurement location Microclimate Soil temperature T s ( C) 3x daily Within 5 15 cm of collar Soil moisture content M s (m 3 /m 3 ) 3x daily Within 5 15 cm of collar Vegetation Basal area BA (m 2 ha 1 ) Winter 2011 15 m radius circular plot from center collar Pine basal area PBA (m 2 ha 1 ) Winter 2011 15 m radiu s circular plot from center collar Hardwood basal area HWBA (m 2 ha 1 ) Winter 2011 15 m radius circular plot from center collar Stand density TPH (trees ha 1 ) Winter 2011 15 m radius circular plot from center collar Distance to the nearest tree Dne arest (m) Winter 2011 Linear distance from soil collar to nearest tree (DBH > 10 cm) Diameter of the nearest tree DBH (cm) Winter 2011 DBH of the nearest tree measured in Dnearest Distance to the nearest palmetto Dist Palm (m) Winter 2011 Linear dista nce from soil collar to center of nearest palmetto Forest floor Duff depth Duff (cm) Winter 2011 Avg. of three measurements within 30 cm of collar Litter depth Litter (cm) Winter 2011 Avg. of three measurements within 30 cm of collar Total duff and l itter depth DL (cm) Winter 2011 Avg. of three measurements within 30 cm of collar Soil organic matter SOM (%) Winter 2011 Avg. of three samples per plot from measurements at 0 5 cm and 5 10 cm depths Soil total carbon TC (%) Winter 2011 Avg. of thr ee samples per plot from measurements at 0 5 cm and 5 10 cm depths Weather Total precipitation Precip (cm) Monthly Osceola records were from the Olustee Remote Automated Weather Station (RAWS #OLSF1). ACMF r ecords were from the Austin Cary Memorial F orest staff onsite rain gauge

PAGE 180

180 Table 4 1. Continued Mean air temperature (2 m) Temp ( C) Monthly Osceola records were from the Olustee Remote Automated Weather Station (RAWS #OLSF1). ACMF records were from the Austin Cary Memorial Forest AmeriFlux To wer. Palmer drought severity index PDSI Monthly North central Florida regional estimate from NOAA NCDC

PAGE 181

181 Table 4 2. Mean forest characteristics per treatment at the Austin Cary Forest and Osceola National Forest, Florida, USA. Site Trt BA (m 2 ha 1 ) T PH (Trees ha 1 ) Duff Depth (cm) Litter Depth (cm) SOM (0 5 cm) (%) SOM (5 10 cm) (%) TC (0 5 cm) (%) TC (5 10 cm) ( %) ACMF 3YR 18.63 (2.43) b 259.35 (32.67) b 0.78 (0.76) b 2.66 (0.91) b 1.32 (0.50) a 0.70 (0.42) a 0.95 (0.61) a 0.55 (1.02) a ACMF 40 YR 27.03 (3.12) a 773.33 (198.22) a 2.87 (1.15) a 4.11 (1.02) a 1.10 (0.31) a 0.50 (0.15) a 0.30 (0.12) a 0.28 (0.81) a Osceola Burn 21.6 3 (1. 51 ) a 4 57 40 ( 94 1 9) a 4.18 ( 0.23 ) a 1.57 (0.21) b 2. 18 (1.26 ) a 0. 74 (0.51 ) a 1.62 (1.03 ) a 0.36 (0.22 ) a Osc eola Control 2 5 08 (3. 73 ) a 5 4 2. 27 (1 55 82 ) a 4.48 (1.94 ) a 4.50 (1.06) a 2.53 (0.42 ) a 0.89 (0.37 ) a 2.86 (2.12 ) a 0.45 (0.12 ) a Osceola Mow 21.9 6 ( 6 14 ) a 330. 08 ( 216.55 ) a 3.27 (2.47 ) a 3.50 (0.51) a 1.32 (0.43 ) a 0.55 (0.21 ) a 0.82 (0.26 ) a 0.22 (0.14 ) a Osceola Mow+ Burn 2 6 04 (2. 89 ) a 3 72 52 ( 94 19 ) a 4.05 (0.44 ) a 1.18 (0.54) b 2.53 (1.01 ) a 1.01 (0.48) a 1.82 (0.80 ) a 0.52 (0.23 ) a Data are means with SD in parentheses. BA is basal area, SOM is soil organic matter content, TC is total soil car bon content.

PAGE 182

182 Table 4 3. Results of the repeated measures ANOVA for soil CO 2 efflux (R s ), soil temperature (T s ), and soil moisture content ( M s ) means for the Austin Cary Memorial Forest and Osceola National Forest Florida, USA R s T s M s Site Analysis p eriod Source df F P > F df F P > F df F P > F ACMF Total Month 11 63.08 < 0.0001* 9 900.65 < 0.0001* 11 45.57 < 0.0001* Treatment*M onth 11 2.43 0.0187* 9 31.09 < 0.0001* 11 3.80 0.0008* Treatment 1 0.35 0.5888 1 5.74 0.0746 1 7.88 0.0484* ACMF Growing Month 7 71.87 < 0.0001* 6 1364.16 < 0.0001* 7 39.74 < 0.0001* Treatment*Month 7 1.12 0.3768 6 39 .34 < 0.0001* 7 2.64 0.0323* Treatment 1 0.68 0.4561 1 11.23 0.0286* 1 8.64 0.0424* ACMF Dormant Month 3 33.21 < 0.0001* 2 313.13 < 0.0001* 3 38.91 < 0.0001* Treatment*Month 3 1.71 0.2180 2 6.36 0.0223* 3 1.42 0.285 5 Treatment 1 11.15 0.0288* 1 0.00 0.9789 1 2.56 0.1851 Osceola Total Month 11 35.69 < 0.0001* 9 321.78 < 0.0001* 11 22.57 < 0.0001* Treatment*Month 33 0.66 0.9123 27 1.30 0.1932 33 1.88 0.0106* Treatment 3 0.86 0 .4985 3 11.42 0.0029* 3 2.33 0.1503 Osceola Growing Month 6 43.05 < 0.0001* 4 551.48 < 0.0001* 6 33.57 < 0.0001 Treatment*Month 18 1.01 0.4666 12 1.72 0.1102 18 2.52 0.0059 Treatment 3 1.21 0.3658 3 31.37 < 0.0001* 3 3.02 0.0939 Osceola Dormant Month 4 35.42 < 0.0001* 4 242.75 < 0.0001* 4 13.35 < 0.0001 Treatment*Month 12 0.20 0.9974 12 0.35 0.9715 12 0.34 0.9753 Treatment 3 0.54 0.6685 3 1.01 0.4355 3 1.30 0.3394 For each month, daily measurements per soil collar were averaged, and the nine soil collar means were then averaged to produce a plot level mean value for each month. The effect of month, treatment, and treatment *month on plot level means were tested for significance (p < 0.05).

PAGE 183

183 Table 4 4. Overall means of soil temperature, moisture content, and soil CO 2 efflux rates per treatment and study site Site Treatment Mean T s (C) Mean M s (m 3 /m 3 ) Mean R s 2 m 2 sec 1 ) ACMF 3YR 20.72 (3.57) a 0.11 (0.08) a 4.12 (1.27) a ACMF 40YR 19.98 (2.69) a 0.07 (0.05) b 4.25 (1.24) a Osceola Burn 19.71 (3.19) a 0.1 2 (0.0 5 ) a 3. 44 (1. 16 ) a Osceola Control 19.14 (2.59) b 0. 09 (0.0 3 ) a 3. 83 (1.07 ) a Osceola M ow 19.75 (3.02) a 0.1 2 (0.0 5 ) a 3. 93 (1. 22 ) a Osceola Mow+Burn 19.92 (3.08) a 0.1 4 (0.0 5 ) a 3. 89 (1. 13 ) a Data are study period means with (SD) per study site and treatment type. Letters indicate significant differences p < 0.05) between treatments per site. Table 4 5. Dormant and growing season mean soil temperature, mean moisture content, and mean soil CO 2 efflux rates per treatment at the Austin Cary Forest and Osceola National Forest, Florida, USA --Dormant se ason ----Growing season --Site Treatment Mean T s (C) Mean M s (m 3 /m 3 ) Mean R s 2 m 2 sec 1 ) Mean T s (C) Mean M s (m 3 /m 3 ) Mean R s 2 m 2 sec 1 ) ACMF 3YR 20.51 (1.44) a 0.05 (0.03) a 3.42 (0.51) b 20.81 (4.19) a 0.14 (0.08) a 4.48 (1.40) a ACMF 40YR 20.50 (1.45) a 0.04 (0.02) a 4.00 (0.67) a 19.76 (3.08) b 0.08 (0.05) b 4.37 (1.44) a Osceola Burn 18.32 (0.78) a 0.12 (0.04) a 3.28 (1.05) a 21.12 (2.62) ab 0.13 (0.05) a 3.04 (0.90) a Osceola Control 18.36 (0.78) a 0.11 (0.03) a 3.73 (1.24) a 19.92 (2.19) c 0.09 (0.03) a 3.55 (0.82) a Osceola Mow 18.64 (0.81) a 0.11 (0.03) a 3.61 (0.96) a 20.78 (2.56) b 0.15 (0.06) a 3.54 (0.61) a Osceola Mow+Burn 18.69 (0.78) a 0.13 (0.04) a 3.81 (1.18) a 21.25 (2.70) a 0.16 (0.06) a 3.35 (0.70) a Data a re means with (SD) per study location and treatment type. Means are for the entire study period per treatment type

PAGE 184

184 Table 4 6. Pe 2 efflux (R s ), soil temperature (T s ), soil moisture content (M s ) and field conditions for the Osceola National Forest Variable R s 2 m 2 sec 1 ) T s (C) M s (m 3 /m 3 ) R s 2 m 2 sec 1 ) 1.00 0.63 0.13 T s (C) 0.63 1.00 0.09 M s (m 3 /m 3 ) 0.13 0.09 1.00 Dist nearest tree (m) 0.06 0.04 0.14 DBH nearest (cm) 0.09 0.03 0.23 Stand density (tree ha 1 ) 0.00 0.03 0.15 Basal area (m 2 ha 1 ) 0.11 0.02 0.03 Dist nearest palmetto (m) 0.18 0.04 0.22 Duff depth (cm) 0.08 0.01 0.12 Litter depth (cm) 0.06 0.09 0.27 Duff+litter depth (cm) 0.09 0.06 0.26 Monthly temp ( C) 0.52 0.77 0.18 Monthly preci p (cm) 0.35 0.28 0.06 Organic matter 0 5 cm (%) 0.01 0.04 0.23 Organic matter 5 10 cm (%) 0.05 0.03 0.12 Correlations are of monthly mean plot measurements with treatments ignored and all treatment x plot x month means pooled. R s is soil CO 2 e fflux 2 m 2 sec 1 ), T s is soil temperature (C), M s is soil volumetric moisture content (m 3 /m 3 ). Plot vegetative and meteorological variables are described further in Table 4 1.

PAGE 185

185 Table 4 CO 2 efflux (R s ), soil temperature (T s ), soil moisture content (M s ) and field conditions for the Austin Cary Forest Variable R s 2 m 2 sec 1 ) T s (C) M s (m 3 /m 3 ) R s 2 m 2 sec 1 ) 1.00 0.89 0.25 T s (C) 0.89 1.00 0.51 M s (m 3 /m 3 ) 0.25 0.50 1.00 Dist nearest tree (m) 0.02 0.10 0.27 DBH nearest (cm) 0.00 0.12 0.19 Stand density (tree ha 1 ) 0.01 0.13 0.25 Basal area (m 2 ha 1 ) 0.03 0.09 0.23 Dist nearest palmetto (m) 0.00 0.01 0.15 Duff depth (cm) 0.02 0.13 0.28 Li tter depth (cm) 0.05 0.11 0.38 Duff+litter depth (cm) 0.03 0.13 0.33 Monthly temp ( C) 0.82 0.80 0.01 Monthly precip (cm) 0.24 0.02 0.40 Organic matter 0 5 cm (%) 0.07 0.07 0.16 Organic matter 5 10 cm (%) 0.11 0.07 0.26 Correlations are of monthly mean plot measurements with treatments ignored and all treatment x plot x month means pooled. R s is soil CO 2 efflux 2 m 2 sec 1 ), T s is soil temperature (C), M s is soil volumetric moisture content (m 3 /m 3 ). Plot vegetative an d meteorological variables are described further in Table 4 1.

PAGE 186

186 Table 4 8. Results of simple linear regression models of soil CO 2 efflux rates and field conditions by study area and treatment Site Treatment Variable Model and estimates R 2 F p ACMF 3YR T s R s = 2.178 + 0.304 *T s 0.83 136.20 < 0.0001 ACMF 3YR Temp R s = 1.067 + 0.262*Temp 0.74 93.63 < 0.0001 ACMF 3YR M s R s = 4.486 3.224*M s 0.04 1.33 0.2565 ACMF 3YR Precip R s = 3.490 + 0.079*Precip 0.08 2.82 0.1026 40YR 40YR T s R s = 4.430 + 0.431*T s 0.85 161.31 < 0.0001 40YR 40YR Temp R s = 0.339 + 0.229*Temp 0.62 56.17 < 0.0001 40YR 40YR M s R s = 4.858 8.748*M s 0.12 4.66 0.0380 40YR 40YR Precip R s = 3.800 + 0.055*Precip 0.04 1.48 0.2315 Osceola Burn T s R s = 0.29 7 + 0.175*T s 0.33 13.96 0.0008 Osceola Burn Temp R s = 1.124 + 0.117*Temp 0.27 12.61 0.0011 Osceola Burn M s R s = 4.451 8.533*M s 0.12 4.49 0.0415 Osceola Burn Precip R s = 2.272 0.157*Precip 0.14 5.61 0.0237 Osceola Burn TPH R s = 6.442 0.007*TPH 0.20 8.30 0.0068 Osceola Burn BA R s = 11.625 0.378*BA 0.17 6.85 0.0131 Osceola Burn Litter R s = 7.712 2.722*Litter 0.17 7.13 0.0116 Osceola Burn Duff+Litter R s = 12.911 1.647*Duff+litter 0.18 7.26 0 .0109 Osceola Burn Dist Palmetto R s = 4.824 1.355*Dist palmetto 0.18 7.39 0.0103 Osceola Control T s R s = 2.009 + 0.295*T s 0.5 5 33.76 < 0.0001 Osceola Control Temp R s = 1.802 + 0.103*Temp 0.24 10.45 0.0028 Osceola Control M s R s = 3.100 + 7.711*M s 0.04 1.53 0.2249 Osceola Control Precip R s = 2.790 + 0.141*Precip 0.13 5.07 0.0311 Osceola Mow T s R s = 0.459 + 0.204*T s 0.58 37.59 < 0.0001 Osceola Mow Temp R s = 1.164 + 0.140*Temp 0.34 17.73 0.0002 Osceola Mow M s R s = 4 .406 3.886*M s 0.03 0.94 0.3391 Osceola Mow Precip R s = 2.610 + 0.177*Precip 0.16 6.49 0.0155

PAGE 187

187 Table 4 8. Continued Osceola Mow+Burn T s R s = 0.226 + 0.191*T s 0.36 15.09 0.0006 Osceola Mow+Burn Temp R s = 1.770 + 0.107*Temp 0.24 10 .66 0.0026 Osceola Mow+Burn M s R s = 4.375 3.537*M s 0.02 0.78 0.3830 Osceola Mow+Burn Precip R s = 3.014 + 0.116*Precip 0.08 2.84 0.1013 For variable descriptions see Table 4 1. Table 4 9 Results of nonlinear models of soil CO 2 efflu x rates using soil temperature as a predictor Site Treatment Model Q 10 R 2 p ACMF 3YR R s = 0.8248 e 0.0761*Ts 2.14 0.80 < 0.001 ACMF 40YR R s = 0.4977 e 0.1049*Ts 2.85 0.82 < 0.001 Osceola Burn R s = 1.1183 e 0.0520*Ts 1.68 0.32 0.001 Osceola Control R s = 0.8319 e 0.0761*Ts 2.14 0.53 < 0.001 Osceola Mow R s = 1.2199 e 0.0539*Ts 1.71 0.56 < 0.001 Osceola Mow+Burn R s = 1.3091 e 0.0500* 1.65 0.35 0.001 Models developed using Equation 4 3 are of monthly mean soil CO 2 efflux rate ( R s ) ( mol CO 2 m 2 sec 1 ) responses to soil temperature (T s ). Coefficients were estimated using statistical software JMP 9.0. Q 10 was calculated using Equation 4 4 (Lundegardh, 1927)

PAGE 188

188 Table 4 10. Results of simple linear regression models of soil CO 2 efflux rates and field conditions by study area, treatment, and season Site Treatment Season Variable Model R 2 F p ACMF 3YR Dormant T s R s = 0.042 + 0.180 *T s 0.50 7.03 0.0328 ACMF 3YR Dormant Temp R s = 1.073 + 0.140*Temp 0.82 48.25 < 0.0001 ACMF 3YR Dormant M s R s = 4 .334 16.630*M s 0.74 28.43 0.0003 ACMF 3YR Dormant Precip R s = 3.970 0.106*Precip 0.81 43.27 < 0.0001 ACMF 3YR Growing T s R s = 2.057 + 0.307*T s 0.89 155.30 < 0.0001 ACMF 3YR Growing Temp R s = 2.473 + 0.326*Temp 0.75 64.58 < 0.0001 ACMF 3Y R Growing M s R s = 5.869 9.594*M s 0.29 8.37 0.0087 ACMF 3YR Growing Precip R s = 3.330 + 0.123*Precip 0.12 2.78 0.1102 ACMF 40YR Dormant T s R s = 2.830 + 0.342*T s 0.53 8.04 0.0252 ACMF 40YR Dormant Temp R s = 1.504 + 0.148*Temp 0.55 12.33 0.0056 ACMF 40YR Dormant M s R s = 4.797 19.870*M s 0.50 9.95 0.0103 ACMF 40YR Dormant Precip R s = 4.671 0.131*Precip 0.73 26.94 0.0004 ACMF 40YR Growing T s R s = 4.650 + 0.448*T s 0.90 174.62 < 0.0001 ACMF 40YR Growing Temp R s = 2.966 + 0.340*Temp 0.81 93.00 < 0.0001 ACMF 40YR Growing M s R s = 5.377 11.898*M s 0.19 5.06 0.0348 ACMF 40YR Growing Precip R s = 2.970 + 0.145*Precip 0.17 4.37 0.0483 Osceola Burn Dormant T s R s = 1.782 + 0.276*T s 0.69 28.57 0.00 01 Osceola Burn Dormant Temp R s = 1.127 + 0.292*Temp 0.35 6.94 0.0206 Osceola Burn Dormant M s R s = 2.282 + 8.766*M s 0.11 1.53 0.2387 Osceola Burn Growing T s R s = 0.941 + 0.1886*T s 0.30 5.65 0.0335 Osceola Burn Growing Temp R s = 1 .922 + 0.237*Temp 0.54 22.70 0.0001 Osceola Burn Growing M s R s = 5.480 15.862*M s 0.44 14.81 0.0011 Osceola Burn Growing TPH R s = 6.703 0.007*TPH 0.19 4.54 0.0464 Osceola Burn Growing BA R s = 12.984 0.436*BA 0.20 4.71 0.042 7 Osceola Burn Growing Litter+duff R s = 14.980 1.994*Litter+Duff 0.23 5.67 0.0279

PAGE 189

189 Table 4 10. Continued Osceola Control Dormant T s R s = 3.293 + 0.382*T s 0.74 37.69 < 0.0001 Osceola Control Dormant Temp R s = 1.844 + 0.368*Temp 0.40 8.7 1 0.0112 Osceola Control Dormant M s R s = 0.506 + 30.079*M s 0.65 15.36 0.0018 Osceola Control Growing T s R s = 1.716 + 0.264*T s 0.51 13.31 0.0030 Osceola Control Growing Temp R s = 0.564 + 0.195*Temp 0.63 30.77 < 0.0001 Osceola Control Gr owing M s R s = 4.857 11.225*M s 0.09 1.73 0.2056 Osceola Control Growing Precip R s = 1.875 + 0.228*Precip 0.34 9.22 0.0071 Osceola Mow Dormant T s R s = 1.623 + 0.281*T s 0.79 44.89 < 0.0001 Osceola Mow Dormant Temp R s = 1.303 + 0.325*Tem p 0.51 13.79 0.0026 Osceola Mow Dormant M s R s = 1.590 + 17.881*M s 0.34 6.79 0.0218 Osceola Mow Growing T s R s = 0.289 + 0.184*T s 0.60 19.27 0.0007 Osceola Mow Growing Temp R s = 1.282 + 0.235*Temp 0.44 15.23 0.0010 Osceola Mow Grow ing M s R s = 5.329 9.022*M s 0.17 3.40 0.0630 Osceola Mow+Burn Dormant T s R s = 2.369 + 0.330*T s 0.70 30.64 < 0.0001 Osceola Mow+Burn Dormant Temp R s = 1.263 + 0.335*Temp 0.37 7.63 0.0161 Osceola Mow+Burn Dormant M s R s = 0.653 + 23.468* M s 0.67 26.22 0.0002 Osceola Mow+Burn Growing T s R s = 0.746 + 0.193*T s 0.55 14.71 0.0024 Osceola Mow+Burn Growing Temp R s = 1.430 + 0.231*Temp 0.66 34.25 < 0.0001 Osceola Mow+Burn Growing M s R s = 6.048 14.954*M s 0.53 20.24 0.0003 O sceola Mow+Burn Growing Precip R s = 1.730 + 0.241*Precip 0.21 4.82 0.0415 For variable descriptions see Table 4 1 (this chapter).

PAGE 190

190 Table 4 11. Results of season specific nonlinear models of the relationships between soil CO 2 efflux rates and soil temperature at the Austin Cary Forest and Osceola National Forest, Florida, USA Site Treatment Season Model Q 10 R 2 p ACMF 3YR Dormant R s = 1.3539 e 0.0482*Ts 1.62 0.49 0.0346 ACMF 3YR Growing R s = 0.9207 e 0.0726*Ts 2.07 0.84 < 0.0001 ACMF 40YR Dorm ant R s = 0.8178 e 0.0792*Ts 2.21 0.51 0.0302 ACMF 40YR Growing R s = 0.4889 e 0.1065*Ts 2.90 0.86 < 0.0001 Osceola Burn Dormant R s = 0.8308 e 0.0737*Ts 2.09 0.64 0.0004 Osceola Burn Growing R s = 0.8731 e 0.0586*Ts 1.80 0.30 0.0352 Osceola Contr ol Dormant R s = 0.6686 e 0.0920*Ts 2.51 0.70 < 0.0001 Osceola Control Growing R s = 0.8671 e 0.0702*Ts 2.02 0.50 0.0034 Osceola Mow Dormant R s = 0.9375 e 0.0713*Ts 2.04 0.76 < 0.0001 Osceola Mow Growing R s = 1.2696 e 0.0489*Ts 1.63 0.58 0.0010 Osc eola Mow+Burn Dormant R s = 0.8687 e 0.0777*Ts 2.17 0.67 0.0002 Osceola Mow+Burn Growing R s = 1.0365 e 0.0548*Ts 1.73 0.55 0.0025 Models developed using Equation 4 3 are of monthly mean soil CO 2 efflux rate ( R s ) ( mol CO 2 m 2 sec 1 ) responses to soi l temperature (T s ). D ata are presented by treatment type, study site, and season. Dormant season defined as October February and growing season defined as March September. Coefficients were estimated using statistical software JMP 9.0. Q 10 was calc ulated using Equation 4 4 (Lundegardh, 1927)

PAGE 191

191 Table 4 12. Step wise multiple linear regression models by study site and treatment predicting soil CO 2 efflux rates from field parameters Site Treatment Equation RMSE R 2 F p ACMF 3YR R s = 1.595 + 0.338*T s + 2.659*M s 2.824*Dist palmetto 0.40 0.90 75.47 < 0.0001 ACMF 40YR R s = 6.782 + 0.526*T s 7.034*M s 0.43 0.89 108.36 < 0.0001 Osceola Burn R s = 2.402 + 0.175*T s 0.006*TPH 0.67 0.56 17.05 < 0.0001 Osceola Control R s = 4.837 + 0.285*T s 11.714*M s + 0.075*BA 0.58 0.72 22.07 < 0.0001 Osceola Mow R s = 1.281 + 0.200*T s + 0.042*BA 0.50 0.65 24.16 < 0.0001 Osceola Mow+Burn R s = 0.226 + 0.191*T s 0.80 0.36 15.09 0.0006 Models developed using a forward step wise procedure with parameter incl usion and retention p < 0.05. For input parameter descriptions see Table 4 2.

PAGE 192

192 Table 4 13. Step wise multiple linear regression models by study site, treatment, and season predicting soil CO 2 efflux rates from field parameters Site Treatment Season Equ ation RMSE R 2 F p ACMF 3YR Dormant R s = 3.970 0.106*Precip 0.23 0.81 43.27 < 0.0001 ACMF 3YR Growing R s = 1.584 + 0.308*T s 0.689*OM (5 10 cm) 0.35 0.94 143.21 < 0.0001 ACMF 40YR Dormant R s = 4.124 0.130*Precip + 0.693*Dist palmetto 0.30 0.83 22.61 0.0003 ACMF 40YR Growing R s = 7.093 + 0.542*T s + 7.249*M s 0.40 0.93 122.39 < 0.0001 Osceola Burn Dormant R s = 0.917 + 0.270*T s 0.006*TPH 0.41 0.87 40.05 < 0.0001 Osceola Burn Growing R s = 1.349 + 0.213*Temp + 0.169*Precip 1.517*Dist pal metto 0.53 0.84 30.57 < 0.0001 Osceola Control Dormant R s = 18.120 + 0.240*T s + 14.331*M s + 0.319*BA + 3.141*Dist tree 0.38 0.93 34.35 < 0.0001 Osceola Control Growing R s = 2.839 + 0.259*T s + 0.146*M s 0.46 0.72 15.74 0.0004 Osceola Mow Dormant R s = 1.966 + 0.269*T s + 0.002*TPH 0.36 0.89 44.69 < 0.0001 Osceola Mow Growing R s = 0.289 + 0.184*T s 0.40 0.60 19.27 0.0007 Osceola Mow +Burn Dormant R s = 27.779 + 0.332*T s 0.906*DBH 0.49 0.85 33.55 < 0.0001 Osceola Mow +Burn Growing R s = 6.048 14.954*M s 0.79 0.53 20.24 0.0003 Models developed using a forward step wise procedure with parameter inclusion and retention p < 0.05. For input parameter descriptions see Table 4 2. Dormant season is October February, Growing season is M arch September.

PAGE 193

193 Figure 4 1. Map of the study areas at the Osceola National Forest near Lake City, Florida and Austin Cary Forest near Gainesville, Florida, USA Map produced by David Godwin.

PAGE 194

194 Figure 4 2. Examples of the four pine flatwoods f orest management types sampled in the Osceola National Forest study site near Lake City, Florida, USA: (A) Unburned control, (B) mow only, (C) burn only, and (D) mow+burn. Mow only understory vegetation regrowth shown is approximately two months following mechanical treatment. Burn only and mow+burn photos depict conditions one day post prescribed burn. Prior to mowing and prescribed fire treatments, conditions within all sites were similar to control units. Photographs courtesy of David Godwin.

PAGE 195

195 Fi gure 4 3. Pine flatwoods forest management types represented in the study at the Austin Cary Memorial Forest, Gainesville, Florida, USA. The prescribed fire sites (A) were burned on a 3 year winter burn rotation while the fire excluded sites (B) were unbu rned for >40 years. Photographs courtesy of David Godwin.

PAGE 196

196 Figure 4 4. Monthly mean soil CO 2 efflux rates (R s 2 m 2 sec 1 ), soil temperature (T s ) (C), and soil moisture content (M s ) (m3/m3) per treatment at the Osceola National Forest near Lake City, Florida, USA. Due to equipment problems T s was not recorded during the months of July and August 2011.

PAGE 197

197 Figure 4 5. Monthly mean soil CO 2 efflux rates (R s 2 sec 1 ), soil temperature (T s ) (C), and soil moisture content (M s ) ( m 3 /m 3 ) per treatment at the Austin Cary Forest near Gainesville, Florida, USA. Due to equipment problems T s was not recorded during the months of July 2010 and January 2011.

PAGE 198

198 Figure 4 6. Monthly Palmer Drought Severity Index values for the region con taining the Austin Cary Forest and the Osceola National Forest study areas. Data are from the National Oceanic and Atmospheric Administration (NOAA) National Climatic Data Center (NCDC). All scores below zero represent increasing levels of regional droug ht severity.

PAGE 199

199 Figure 4 7. Treatment means of soil CO 2 efflux rates (R s 2 m 2 sec 1 ), soil temperature (T s ) (C) and soil volumetric moisture content (M s ) (m 3 /m 3 ) for the Osceola National Forest during the sampling period of March 2011 March Figure 4 8. Treatment means of soil CO 2 efflux rates (R s 2 m 2 sec 1 ), soil temperature (T s ) (C) and soil volumetric moisture content (M s ) (m 3 /m 3 ) fo r the Austin Cary Forest during the sampling period of February 2010 June 2011. test p < 0.05).

PAGE 200

200 Figure 4 9. Linear regressions of the relationships between mean soil CO 2 efflu x rates (R s 2 m 2 sec 1 ) and mean: soil temperature (C), monthly mean air temperature (C), soil moisture content (m 3 /m 3 ), monthly total precipitation (cm), basal area (m 2 ha 1 ), stand density (tree ha 1 ), distance to nearest tree (m), and distance to neare st palmetto (m) for the Osceola National Forest. Each point represents entire study period means per sample plot with all treatments combined.

PAGE 201

201 Figure 4 10. Linear regression of the relationships between mean soil CO 2 efflux rates (R s 2 m 2 se c 1 ) and mean: duff depth (cm), litter depth (cm), and litter+duff depth (cm) for the Osceola National Forest. Each point represents entire study period means per sample plot with all treatments combined.

PAGE 202

202 Figure 4 11 Linear regression of the relat ionships between mean soil CO 2 efflux rates (R s 2 m 2 sec 1 ) and mean: soil temperature (C), mean monthly air temperature (C), soil moisture content (m 3 /m 3 ), total monthly precipitation (cm) basal area (m 2 ha 1 ), stand density (tree ha 1 ), distance to nearest tree (m), and distance to neare st palmetto (m) for the Austin Cary Forest Each point represents entire study period means per sample plot with all treatments combined.

PAGE 203

203 Figure 4 12 Linear regression of the relationships between mean soil CO 2 efflux rates (R s 2 m 2 sec 1 ) and mean: duff depth (cm), litter depth ( cm), and litter+duff depth (cm) for the Austin Cary Forest Each point represents entire study period means per sample plot with all treatments combined.

PAGE 204

204 Figure 4 1 3 Linear regression of the relationships b etween monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and soil temperature (T s ) ( C) (top four plots) as well as R s and monthly mean air temperature ( M Temp ) ( C) (bottom four plots) for treatments at the Osceola National Forest Each point represents monthly mean values per sample plo t.

PAGE 205

205 Figure 4 1 4 Linear regression of the relationships between monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and soil temperature (T s ) ( C) (top two plots) as well as R s and monthly mean air temperature ( M Temp ) ( C) (bottom to plots) fo r treatments at the Austin Cary Forest Each point represents monthly mean values per sample plot.

PAGE 206

206 Figure 4 1 5. Non l inear regression of the relationships between monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and soil temperature (T s ) ( C) (top four plots) as well as R s and monthly mean air temperature ( M Temp ) ( C) (bottom four plots) for treatments at the Osceola National Forest Each point represents monthly mean values per sample plot.

PAGE 207

207 Figure 4 1 6. Non l inear regression of the relationships between monthly mean soil CO 2 efflux rates ( R s 2 m 2 sec 1 ) and soil temperature (T s ) ( C) (top two plots) as well as R s and monthly mean air temperature ( M Temp ) ( C) (bottom two plots) for treatments at the Austin Cary Forest Each point represents monthly mean values per sample plot.

PAGE 208

208 Figure 4 1 7 Seasonal (dormant and growing) l inear regression s of the relationships between monthly mean soil CO 2 efflux rates (R s 2 m 2 sec 1 ) and soil temperature (T s ) (C) for treatments at the Osceola National Forest. Each point represents monthly mean values per sample plot.

PAGE 209

209 Figure 4 1 8 Seasonal (dormant and growing) non l inear regression s of the relationships between monthly mean soil CO 2 efflux rates (R s 2 m 2 sec 1 ) and soil temperature (T s ) (C) per treatment at the Osce ola National Forest. Each point represents monthly mean values per sample plot.

PAGE 210

210 Figure 4 1 9 Seasonal (dormant and growing) l inear regression s of the relationships between monthly mean soil CO 2 efflux rates (R s 2 m 2 sec 1 ) and soil temperat ure (T s ) (C) per treatment at the Austin Cary Forest. Each point represents monthly mean values per sample plot.

PAGE 211

211 Figure 4 20 Seasonal (dormant and growing) non l inear regression s of the relationships between monthly mean soil CO 2 efflux rates (R s ) 2 m 2 sec 1 ) and soil temperature (T s ) (C) per treatment at the Austin Cary Forest. Each point represents monthly mean values per sample plot.

PAGE 212

212 Figure 4 2 1 Predicted monthly total soil carbon flux (g C m 2 month 1 ) for the four treatment s at the Osceola National Forest near Lake City, Florida, USA for the period March 2011 February 2012. Flux values were predicted using treatment specific linear models of soil CO 2 efflux response to changes in soil temperature. Model input hourly mean 10 cm depth soil temperature for the period of March 2011 February 2012 was recorded at the nearby Mac c lenny, Florida Automated Weather Network (FAWN) station.

PAGE 213

213 Figure 4 2 2 Predicted annual total soil carbon flux (kg C m 2 yr 1 ) for the four treatment s at the Osceola National Forest near Lake City, Florida, USA for the period of March 2011 February 2012. Flux values were predicted using treatment specific linear models of soil CO 2 efflux response to changes in 10 cm soil temperature. Model input ho urly mean soil temperature was recorded at the nearby Mac cl enny, Florida Automated Weather Network (FAWN) station.

PAGE 214

214 Figure 4 23. Predicted monthly total soil carbon flux (g C m 2 month 1 ) for the two prescribed fire treatments at the Austin Cary Forest, Gainesville, Florida, USA for the period of March 2010 February 2011. Flux values were predicted using treatment specific linear models of soil CO 2 efflux response to changes in soil te mperature. Model input hourly mean 10 cm depth soil temperature recorded at the Putnam Hall, Florida, FAWNS station.

PAGE 215

215 Figure 4 24. Predicted annual total soil carbon flux (kg C m 2 yr 1 ) for the period of March 2010 February 2011 for two prescribed fire treatments at the Austin Cary Forest near Gainesville, Florida, USA. Flux values were predicted using treatment specific linear models of soil CO 2 efflux response to changes in 10 cm soil temperature. Model input hourly mean soil temperature record ed at the Putnam Hall, Florida Automated Weather Network (FAWN) station.

PAGE 216

216 CHAPTER 5 SUMMARY AND SYNTHESIS Summary The results of the studies presented here expand the understanding of the influence of prescribed fire and mechani cal fuels mastication treatments on soil CO 2 efflux rates in Florida old field and flatwoods forests. In Chapter 2, the results of a nearly two year study of soil CO 2 efflux rates at Tall Timbers Research Station found differences in rates among three diff erent prescribed fire management regimes. It was found that prolonged management regimes utilizing frequent prescribed fire in loblolly pine shortleaf pine old field forests drive large shifts in forest structure and composition that result in reduced m onthly mean soil CO 2 efflux rates as compared to a management regime of prolonged fire exclusion. Average monthly mean soil CO 2 efflux rates in the annually burned forests were approximately 37% lower than those in the long unburned forests. In addition, estimated annual soil carbon fluxes based on the response of monthly diurnal soil CO 2 efflux rates to changes in soil temperature, found that total annual soil carbon efflux was lower in the annually (1069 g m 2 y 1 ) and biennially (1268 g m 2 y 1 ) burned forests than in the unburned forest (1688 g m 2 y 1 ). In Chapter 3, a seven month litter manipulation experiment in a loblolly pine shortleaf pine old field forest supported the conclusions of Chapter 2: frequent prescribed burning reduces soil CO 2 efflu x rates relative to fire exclusion, with annual burning resulting in lower soil CO 2 efflux rates than biennial burning. In addition, soil CO 2 efflux rates in frequently burned old field forests respond positively to one time elevated litter inputs, with h igher rates persisting for at least several months following

PAGE 217

217 litter additions. Soil CO 2 efflux rates in the same forests do not appear sensitive to short term reductions in leaf litter inputs, although longer term studies may yet identify seasonal relatio nships between litter inputs and soil CO 2 efflux rates. The results of this study indicate that prolonged prescribed fire management regimes result in changes in the importance of aboveground litter inputs on the heterotrophic sources of soil CO 2 efflux, with frequently burned sites, rather than long fire excluded sites, much more sensitive to increases in litterfall. In Chapter 4, a study in two mature flatwoods forests found that neither prescribed fire nor mechanical fuels mastication treatments nor me chanical fuels mastication followed by prescribed fire significantly influenced mean soil CO 2 efflux rates. Prescribed fire and mechanical fuels mastication treatments at the Osceola National Forest study site were found to significantly increase monthly mean soil temperature. At the Austin Cary study site, a management regime of prescribed fire was shown to increase soil moisture content relative to a management regime of fire exclusion. These results, while indicating no direct influences on soil CO 2 e fflux rates, suggest that prolonged periods of continuous management may affect soil carbon dynamics through persistent changes in the abiotic conditions that influence heterotrophic and autotrophic sources of soil carbon efflux. Further research is neede d to understand the long term implications of increased soil temperature in these systems, particularly in sites where mechanical fuels mastication treatments have occurred, as the fate and duration of the ecosystem effects of masticated fuel treatments ar e not well known. In Chapters 2, 3, and 4, soil CO 2 efflux rates were generally strongly positively correlated with soil temperature and monthly mean ambient air temperature. These

PAGE 218

218 results are consistent with many previous studies of soil CO 2 efflux rates in multiple ecosystems (Raich and Schlesinger, 1992; Ryan and Law, 2005; Luo and Zhou, 2006). Given this correlation and the magnitude of total global soil carbon emissions ( 75 Pg C yr 1 ), our results provide evidence of a need for continued research reg arding the influence of changing global temperatures on soil CO 2 efflux rates. Chapters 2, 3, and 4 all provided some non conclusive evidence suggesting that seasonal variations, as well as prescribed fire and mechanical fuels mastication treatments, may i nfluence the relative contributions of heterotrophic and autotrophic sources of soil CO 2 efflux. To better predict the effects of global climate change on soil CO 2 efflux rates, more research is needed to understand how autotrophic and heterotrophic sourc es of CO 2 respond to forest management practices, as well as to changes in elevated atmospheric CO 2 concentrations, temperature, moisture regimes, and forest vegetation. Our results suggest that future studies would benefit from both short and long term s ampling intervals that capture the daily, monthly, and seasonal variability in soil CO 2 efflux rates. In addition, cost effective methods that allow for the partitioning of the components of heterotrophic and autotrophic sources of soil CO 2 efflux in situ without disturbing site biotic or abiotic factors are greatly needed. The research presented in these studies provides evidence suggesting that future models of southeastern US forest soil CO 2 efflux rates account for forest type specific responses to man agement practices. In Chapters 2 and 3, a management regime of frequent prescribed fire in old field forests resulted in conditions that supported lower soil CO 2 efflux rates and soil carbon emissions. In contrast, in Chapter 4, a prolonged prescribed fi re management program at the Austin Cary flatwoods site did not

PAGE 219

219 significantly influence soil CO 2 efflux rates. While these studies provide insight into management influences on soil CO 2 efflux rates, it is important to remember that soil CO 2 efflux is onl y one component of the complex forest carbon cycle (Wardle et al., 2004). Additional research that assesses direct carbon emissions from prescribed fires, as well subsequent post burn vegetative responses is needed to fully understand the implications of long term prescribed fire and mechanical fuels mastication management programs on total ecosystem carbon dynamics and budgets. Synthesis In the studies reported here it was predicted that a frequent pre scribed fire management regime would result i n reduced soil respiration rates relative to a management regime of fire exclusion. This hypothesis was based primarily on previous research describing the influence of prescribed fire on forest biomass, composition, and litter pools (Glitzenstein et al., 2012; Lavoie et al., 2010; Reid et al., 2012) and previous research describing the importance of those and similar factors in driving the autotrophic and heterotrophic sources of soil respiration rates (Ryan and Law, 2005; Kuzyakov, 2006; Lou and Zhou, 20 06; Sulzman et al., 2012). The research of those authors and others was used to develop a conceptual model that identified a wide range of biotic and abiotic factors known to influence autotrophic, heterotrophic, and total soil respiration rates (Figure 5 1). As soil respiration is the sum of heterotrophic respiration from soil microbial metabolism and autotrophic respiration from live root and rhizosphere fungi activity, it was hypothesized that changes in aboveground vegetative characteristics caused by prescribed fire would have significant impacts on total measured soil respiration rates. In addition, it was hypothesized that mechanical fuels mastication treatments, due to their impact on aboveground vegetative biomass,

PAGE 220

220 structure, and forest floor cha racteristics, would also reduce soil respiration rates relative to untreated sites. To test these hypotheses, studies were established in loblolly pine shortleaf pine old field sites and longleaf pine slash pine flatwoods sites managed with prescribed fire and mechanical fuels mastication (flatwoods only). Following over three and a half years of measurement at multiple study sites, relationships were identified between soil respiration rates and forest management methods, vegetative characteristics, a nd abiotic conditions. To illustrate the influence of prescribed fire on soil respiration rates, a more specific conceptual model was developed based on the results of the studies reported here (Figure 5 2). The research reported here demonstrated that in the old field forests, a forest management regime utilizing frequent prescribed fire, when maintained for an extended period of time, can result in significant shifts in ecosystem structure and composition as compared to a management regime of fire exclusion. It was determined that prescribed fire frequency can alter certain site biotic characteristics resulting in lower soil respiration rates. The results of these studies found that increasing prescribed fire frequency drives lower total abovegro und living biomass, hardwood abundance relative to conifer abundance, and duff and litter accumulation (Figure 5 2) Reductions in aboveground living biomass were generally associated with lower soil respiration rates, likely due to lower autotrophic and heterotrophic soil respiration. Reductions in hardwood vegetative abundance were also generally associated wi th lower soil respiration rates. This was likely due to reductions in the quantity of deciduous hardwood leaf litter relative to conifer litter. Hardwood litter typically has higher nutrient content than conifer litter and the quality of leaf litter has been shown to positively influence heterotrophic soil

PAGE 221

221 respiration rates in previous studies (Luo and Zhou, 2006). Reductions in forest floor duff and litter accumulation likely resulted in lower heterotrophic soil respiration rates, as overall soil respiration rates were found in frequently burned sites to respond positively to increases in aboveground litter inputs. As none of the studies reporte d here attempted to explicitly partition the sources of soil respiration, we can only infer the specific responses of autotrophic and heterotrophic sources of respiration based on the results of the litter manipulations and previous studies in the literatu re. The studies reported here also documented the importance of season, weather, and management activities on soil temperature and moisture content (Figure 5 2) Prescribed fire frequency and mechanical fuels mastication treatments were shown to influence soil temperature and soil moisture, largely due to changes associated with forest vegetative cover and forest floor exposure. While soil temperature and to a much lesser extent soil moisture, were shown to influence temporal variations in soil respiratio n rates, they did not explain differences in soil respiration rates among management regimes. This suggests that while soil temperature and soil moisture content are important factors influencing photosynthesis and belowground carbon allocation by plants and enzymatic activity by soil microbes, it is the effect of management activities on biotic characteristics that drives differences in overall soil respiration rates between different sites In contrast to the initial hypothesis, the results did not find mechanical fuels mastication treatments and prescribed fire frequency to reduce soil respiration rates in flatwoods sites. It is possible that in the flatwoods sites neither prescribed fire nor mechanical fuels mastication treatment frequency or mana gement regime tenure were

PAGE 222

222 sufficient to drive changes in site biotic or abiotic characteristics that would result in changes in total soil respiration rates. It is also possible that treatments in the flatwoods sites may have induced compensatory shifts i n the response of autotrophic and heterotroph ic sources of soil respiration that masked changes in overall soil respiration rates.

PAGE 223

223 Figure 5 1. Conceptual model of the sources and drivers of soil respiration rates in forested ecosystems managed with prescribed fire Depicted drivers and sources were determined through a survey of previously published research. Illustration by David Godwin.

PAGE 224

224 Figure 5 2. Conceptual model of the influence of prescribed fire frequency and seasonal and environmental factors on soil respiration rates Red italicized text indicates relationships among factors that were quantified in the studies reported in this document Black italicized text indicates possible relationships t hat were supported by the results of previous studies in the literature. Illustration by David Godwin

PAGE 225

225 LIST OF REFERENCES Agee, J.K., Skinner, C.N., 2005. Basic principles of forest fuel reduction treatments. For. Ecol. Manage. 211, 83 96. Alameda, D., footprint on soil physical properties. For. Ecol. Manage. 283, 128 137. Amiro, B.D., Barr, a. G., Barr, J.G., Black, T. a., Bracho, R., Brown, M., Chen, J., Clark, K.L., Davis, K.J ., Desai, a. R., Dore, S., Engel, V., Fuentes, J.D., Goldstein, a. H., Goulden, M.L., Kolb, T.E., Lavigne, M.B., Law, B.E., Margolis, H. a., Martin, T., McCaughey, J.H., Misson, L., Montes Helu, M., Noormets, a., Randerson, J.T., Starr, G., Xiao, J., 2010. Ecosystem carbon dioxide fluxes after disturbance in forests of Nor th America. J. Geophys Res. 115, G00K02. Baggs, E., 2006. Partitioning the components of soil respiration: a research challenge. Plant Soil 284, 1 5. Bellon, A., Austin, G.L., 1986. On the relative accuracy of satellite and rain ga u ge rainfall measurements over middle latitudes during daylight hours. J. Appl. Meteorol. 25, 1712 1724. Bhupinderpal Singh, Nordgren, A., Ottosson Lofvenius, M., Hogberg, M.N., Mellander, P.E., Hogberg, P., 2003. Tree root and soil heterotrophic respiration as revealed by girdling of boreal Scots pine forest: extending observations beyond the first year. Plant Cell Environ. 26, 1287 1296. Blagodatskaya, E., Yuyukina, T., Blagodatsky, S., Kuzyakov, Y., 2011. Three source partitioning of microbial biomass and of CO 2 efflux from soil to evaluate mechanisms of priming effects. Soil Biol. Biochem. 1 9. Boerner, R.E.J., Waldrop, T.A., Shelburne, V.B., 2006. Wildfire mitigation strategies affect soil enzyme activity and s oil organic carbon in loblolly pine (Pinus taeda) forests. Can. J. For. Res. 36, 3148 3154. Bonan, G.B., 2008. Forests and climate change: forcings, feedbacks, and the climate benefits of forests. Science 320, 1444 9. Bond Lamberty, B., Thomson, A., 2010. Temperature associated increases in the global soil respiration record. Nature 464, 579 582. Boone, R., Nadelhoffer, K., Canary, J., Kaye, J., 1998. Roots exert a strong influence on the temperature sensitivity of soil respiration. Nature 396, 570 572. Bow den, R., Nadelhoffer, K., Boone, R., Melilo, J., Garrison, J., 1993. Contributions of aboveground litter, belowground litter, and root respiration to total soil respiration in a temperate mixed hardwood forest. Can. J. For. Res. 23, 1402 1407.

PAGE 226

226 Bracho, R., Starr, G., Gholz, H., Martin, T., Cropper, W., Loescher, H., 2012. Controls on carbon dynamics by ecosystem structure and climate for southeastern U.S. slash pine plantations. Ecol. Monogr. 82, 101 128. Brose, P., Wade, D., 2002. Potential fire behavior in pine flatwood forests following three different fuel reduction techniques. For. Ecol. Manage. 163, 71 84. Burger, J.A., Pritchett, W.L., 1988. Site p reparation e ffects on s oil m oisture. For. Sci. 34, 77 87. Butnor, J., Johnsen, K., Oren, R., Katul, G., 20 03. Reduction of forest floor respiration by fertilization on both carbon dioxide enriched and reference 17 year old loblolly pine stands. Glob. Change Biol. 9, 849 861. Carney, K.M., Hungate, B. a, Drake, B.G., Megonigal, J.P., 2007. Altered soil microbia l community at elevated CO 2 leads to loss of soil carbon. Proceedings of the National Academy of Sciences of the United States of America 104, 4990 5. Castro, M.S., Gholz, H .L., Clark, K.L., Steudler, P.A ., 2000. Effects of forest harvesting on soil methan e fluxes in Florida slash pine plantations. Can. J. For. Res. 1534 1542. Certini, G., 2005. Effects of fire on properties of forest soils: a review. Acta Oecol. 143, 1 10. Chemidlin Prvost Bour, N., Soudani, K., Damesin, C., Berveille r, D., Lata, J. Duf rne, E., 2010. Increase in aboveground fresh litter quantity over stimulates soil respiration in a temperate deciduous forest. Appl. Soil Ecol. 46, 26 34. Choromanska, U., DeLuca, T.H., 2002. Microbial activity and nitrogen mineralization in forest minera l soils following heating: evaluation of post fire effects. Soil Biol. Biochem. 34, 263 271. Cisneros Dozal, L.M., Trumbore, S.E., Hanson, P.J., 2007. Effect of moisture on leaf litter decomposition and its contribution to soil respiration in a temperate f orest. J. Geophys Res. 112, 1 10. Clark, K., Gholz, H. Castro, M.S., 2004. Carbon dynamics along a chronosequence of slash pine plantations in North Florida. Ecol. Appl. 14, 1154 1171. Cleveland, C.C., Nemergut, D.R., Schmidt, S.K., Townsend, A.R., 2006. In creases in soil respiration following labile carbon additions linked to rapid shifts in soil microbial community composition. Biogeochemistry 82, 229 240. Clewell, A., Komarek, R., 1975. NB66: The initiation of a long term experiment in forest succession. Unpublished manuscript.

PAGE 227

227 Clinton, B.D., Maier, C ., Ford, C.R., Mitchell, R.J., 2011. Transient changes in transpiration, and stem and soil CO 2 efflux in longleaf pine (Pinus palustris Mill.) following fire ind uced leaf area reduction. Trees 25, 997 1007. Ci sneros Dozal, L.M., Trumbore, S.E., Hanson, P.J., 2007. Effect of moisture on leaf litter decomposition and its contribution to soil respiration in a temperate forest. J. Geophys. Res. 112, 1 10. Concilio, A., Ma, S., Li, Q., LeMoine, J., Chen, J., North, M., Moorhead, D., Jensen, R., 2005. Soil respiration in response to prescribed burning and thinning in mixed conifer and hardwood forests. Can. J. For. Res. 35, 1581 1591. Cropper, W., Gholz, H., 1991. In situ needle and fine root respiration in mature sla sh pine (Pinus elliottii) trees. Can. J. For. Res. 21, 1589 1595. Davidson, E., Richardson, A., Savage, K., Hollinger, D., 2006. A distinct seasonal pattern of the ratio of soil respiration to total ecosystem respiration in a spruce dominated forest. Glob. Change Biol. 12, 230 239. Debano, L., 2000. The role of fire and soil heating on water repellency in wildland environments: a review. J. Hydrol. 231 232, 195 206. Ekblad, A., Hogberg, P., 2001. Natural abundance of 13C in CO2 respired from forest soils re veals speed of link between tree photosynthesis and root respiration. Acta Oecol. 127, 305 308. Engstrom, R.T., Palmer, W.E., 2005. Two species in one ecosystem: management of Northern bobwhite and Red cockaded woodpecker in the Red Hills, in: Ralph, J., R ich, T. (Eds.), Bird Conservation Implementation and Integration in the Americas: Proceedings of the Third International Partners in Flight Conference. USDA Forest Service Pacific Southwest Research Station, Albany, CA, pp. 1151 1157. Epron, D., Bahn, M., Derrien, D., Lattanzi, F.A., Pumpanen, J., Gessler, A., Hgberg, P., Maillard, P., Dannoura, M., Grant, D., Buchmann, N., 2012. Pulse labeling trees to study carbon allocation dynamics: a review of methods, current knowledge and future prospects. Tree Phy sio. 32, 776 98. Ewel, K., Cropper, W., Gholz, H., 1987a. Soil CO 2 evolution in Florida slash pine plantations. I. Changes through time. Can. J. For. Res. 17, 325 329. Ewel, K., Cropper, W., Gholz, H. 1987b. Soil CO 2 evolution in Florida slash pine planta tions II. Importance of root respiration. Can. J. For. Res. 17, 330 333. Fang, C., Moncrieff, J., Gholz, H., Clark, K 1998. Soil CO 2 efflux and its spatial variation in a Florida slash pine plantation. Plant and Soil 135 146.

PAGE 228

228 Fenn, K.M., Malhi, Y., Morec roft, M.D., 2010. Soil CO 2 efflux in a temperate deciduous forest: Environmental drivers and component contributions. Soil Biol. Biochem. 42, 1685 1693. Frost, C., 1993. Four centuries of changing landscape patterns in the longleaf pine ecosystem, in: Herm ann, S.M. (Ed.), Proceedings 18th Tall Timbers Fire Ecology Conference. The Longleaf Pine Ecosystem: Ecology, Restoration and Management. Tall Timbers Research, Inc., Tallahassee, FL, pp. 17 43. Garten Jr., C.T., 2009. A disconnect between O horizon and mi neral soil carbon Implications for soil C sequestration. Acta Oecologica 35, 218 226. Gholz, H., Fisher, R., 1982. Organic matter production and distribution in slash pine (Pinus elliottii ) plantations. Ecology 63, 1827 1839. Gholz, H., Clark, K.L., 200 2. Energy exchange across a chronosequence of slash pine forests in Florida. Agr. Forest Meteorol. 112, 87 102. Glitzenstein, J., Streng, D., Achtemeier, G.L., Naeher, L.P., Wade, D.D., 2006. Fuels and fire behavior in chipped and unchipped plots: Implicat ions for land management near the wildland/urban interface. For. Ecol. Manage. 236, 18 29. Glitzenstein, J., Streng, D. Masters, R.E., Robertson, K.M., Hermann, S.M., 2012. Fire frequency effects on vegetation in north Florida pinelands: Another look at t he long term Stoddard Fire Research Plots at Tall Timbers Research Station. For. Ecol. Manage. 264, 197 209. Gough, C.M., Seiler, J.R., 2004. The influence of environmental, soil carbon, root, and stand characteristics on soil CO 2 efflux in loblolly pine ( Pinus taeda L.) plantations located on the South Carolina Coastal Plain. For. Ecol. Manage. 191, 353 363. Gough, C.M., Seiler, J .R., Wiseman, P.E., Maier, C. 2005. Soil CO 2 efflux in loblolly pine (Pinus taeda L.) plantations on the Virginia Piedmont and South Carolina Coastal Plain over a rotation length chronosequence. Biogeochemistry 73, 127 147. Hanson, P.J., Edwards, N.T., Garten, C.T., Andrews, J.A., 2000. Separating root and soil microbial contributions to soil respiration: a review of methods and o bservations. Biogeochemistry 48, 115 146. Hanula, J.L., Ulyshen, M.D., Wade, D.D., 2012. Impacts of prescribed fire frequency on coarse woody debris volume, decomposition and termite activity in the longleaf pine flatwoods of Florida. Forests 3, 317 331. H einemeyer, A., McNamara, N.P., 2011. Comparing the closed static versus the closed dynamic chamber flux methodology: Implications for soil respiration studies. Plant Soil 346, 145 151.

PAGE 229

229 Hinesley, L.E., Nelson, L.E., 1991. Weight and nutrient content of litt er during secondary succession on well drained uplands of the East Gulf Coastal Plain in Mississippi. Can. J. For. Res. 21, 848 857. Hoosbeek, M.R., 2004. More new carbon in the mineral soil of a poplar plantation under Free Air Carbon Enrichment (POPFACE) : c ause of increased primi ng effect? Global Biogeochem. Cy. 18, 1 7. Hurteau, M., North, M., 2009. Fuel treatment effects on tree based forest carbon storage and emissions under modeled wildfire scenarios. Front. Ecol. Environ. 7, 409 414. IPCC, 1995. Scie ntific Assessments of Climate Change. of Working Group 1 to the Intergovernmental Panel on Climate Change. WMO/UNEP. Johnson, D., 2001. Effects of forest management on soil C and N storage: meta analysis. For. Ecol. Manage. 140, 227 238. Johnson, K.H., Wear, D., Oren, R., Teskey, R.O., Sanchez, F., Seiler, J., Ellsworth, D., Maier, C., Katul, G., Dougherty, P.M., 2001. Meeting global policy commitments: Carbon sequestration and southern pine forests. J. For. 99, 14 21. Johnson, D. W., Knoepp, J.D., Swank, W.T., Shan, J., Morris, L., Van Lear, D.H., Kapeluck, P.R., 2002. Effects of forest management on soil carbon: results of some long term resampling studies. Eviron. Pollut. 116, S201 8. Jonasson, S., Castro, J., Michelsen, A., 2004 Litter, warming and plants affect respiration and allocation of soil microbial and plant C, N and P in arctic mesocosms. Soil Biol. Biochem. 36, 1129 1139. Cisneros Dozal, L.M., Trumbore, S.E., Hanson, P.J., 2007. Effect of moisture on leaf litter decomp osition and its contribution to soil respiration in a temperate forest. Geophys Res. 112, G01013. Karl, T., Melilo, J., Peterson, T., 2009. Global Climate Change Impacts in the United States. Cambridge University Press, New York, NY. Katayama, A., Kume, T., Komatsu, H., Ohashi, M., Nakagawa, M., Yamashita, M., Otsuki, K., Suzuki, M., Kumagai, T., 2009. Effect of forest structure on the spatial variation in soil respiration in a Bornean tropical rainforest. Agr. Forest Meteorol. 149, 1666 1673. Knapp, E., Est es, B., Skinner, C. 2009. Ecological effects of prescribed fire season: a literature review and synthesis for managers. Gen. Tech. Rep. PSW GTR 224. US Department of Agriculture, Forest Service, Pacific Southwest Research Station, Albany CA 80 p.

PAGE 230

230 Kobziar L.N., 2007. The role of environmental factors and tree injuries in soil carbon respiration response to fire and fuels treatments in pine plantations. Biogeochemistry 84, 191 206. Kobziar, L.N., McBride, J.R., Stephens, S.L., 2009. The efficacy of fire an d fuels reduction treatments in a Sierra Nevada pine plantation. Int. J. Wildland Fire 18, 791. Kobziar, L.N., Stephens, S.L., 2006. The effects of fuels treatments on soil carbon respiration in a Sierra Nevada pine plantation. Agr. Forest Meteorol. 141, 1 61 178. Kreye, J., 2012. Efficacy and ecological effects of mechanical fuel treatments in pine flatwoods ecosystems of Florida, USA. Ph.D. Dissertation, University of Florida, Gainesville, Florida. p. 185 Kutsch, W.L., Persson, T., Schrumpf, M., Moyano, F. E., Mund, M., Andersson, S., Schulze, E. D., 2010. Heterotrophic soil respiration and soil carbon dynamics in the deciduous Hainich forest obtained by three approaches. Biogeochemistry 100, 167 183. Kuzyakov, Y., 2006. Sources of CO 2 efflux from soil and r eview of partitioning methods. Soil Biol. Biochem. 38, 425 448. Kuzyakov, Y., 2010. Priming effects: Interactions between living and dead organic matter. Soil Biol. Biochem. 42, 1363 1371. Lal, R., 2005. Forest soils and carbon sequestration. For. Ecol. Ma nage. 220, 242 258. Lau, J., Lennon, J.T., 2012. Rapid responses of soil microorganisms improve plant fitness in novel environments. PNAS. 109, 14058 62. Lavoie, M., Star r, G., Mack, M.C., Martin, T. Gholz, H.L., 2010. Effects of a p rescribed f ire on u nde rstory v egetation, c arbon p ools, and s oil n utrients in a l ongleaf p ine s lash p ine f orest in Florida. Nat. Area J 30, 82 94. Law, B.E., Harmon, M.E., 2011. Forest sector carbon management measurement and verification and discussion of policy related to m itigation and adaptation of forests to climate change. Carbon Manag. 2, 1 12. Law, B.E., Sun, O.J., Campbell, J., Van Tuyl, S., Thornton, P.E., 2003. Changes in carbon storage and fluxes in a chronosequence of ponderosa pine. Glob. Change Biol. 9, 510 524. Lee, M Nakane, K., Nakatsubo, T., Koizumi, H., 2003. Seasonal changes in the contribution of root respiration to total soil respiration in a cool temperate deciduous forest. Plant Soil 255, 311 318.

PAGE 231

231 Li, C.H., Ma, B.L., Zhang, T.Q., 2012. Soil bulk densit y effects on soil microbial populations and enzyme activiti es during the growth of maize ( Zea mays L .) planted in large pots under field exposure. Can. J. Soil Sci. 82, 147 154. Li, Y., Xu, M., Sun, O., Cui, W., 2004. Effects of root and litter exclusion on soil CO 2 efflux and microbial biomass in wet tropical forests. Soil Biol. Biochem. 36, 2111 2114. Liu, Y., Stanturf, J., Goodrick, S., 2010. Trends in global wildfire potential in a changing climate. For. Ecol. Manage. 259, 685 697. Long, A.J., Wade, D. Beall, F.C., 2004. Managing for fire in the interface: challenges and opportunities, in: Vince, S.W., Duryea, M.L., Macie, E.A., Hermansen, L.A. (Eds.), Forests at the Wildland urban Interface. CRC Press, Boca Raton, FL, pp. 201 223. Luan, J., Liu, S., W ang, J., Zhu, X., Shi, Z., 2011. Rhizospheric and heterotrophic respiration of a warm temperate oak chronosequence in China. Soil Biol. Biochem. 43, 503 512. Lundegardh, H., 1927. Carbon dioxide evolution of soil and crop growth. Soil Sci. 23, 417 453. Luo Y., Zhou, X., 2006. Soil Respiration and the Environment. Academic Press, Burlington, MA. Luyssaert, S., Schulze, D., Brner, A., Knohl, A., Hessenmller, D., Law, B.E., Ciais, P., Grace, J., 2008. Old growth forests as global carbon sinks. Nature 455, 2 13 5. Madsen, R.A., Xu, I., Anderson, D.J., Demetrades Shah, T.H., Garcia, R.L., McDermitt, D.K., 2008. Soil CO2 flux measurements LI 6400 and LI 8100 (Research Poster). Lincoln, NE. Maier, C.A., Albaugh, T.J., Allen, H .L., Dougherty, P.M., 2004. Respiratory carbon use and carbon storage in mid rotation loblolly pine (Pinus taeda L.) plantations: the effect of site resources on the stand carbon balance. Glob. Change Biol. 10, 1335 1350. Maier, C. A. Johnsen, K., Butnor, J., Nelson, D., 2012. Southern f orest s cience in s upport of a l ow c arbon e conomy. For. Sci. 58, 2 2. Maier, C.A., Kress, L.W., 2000. Soil CO 2 evolution and root respiration in 11 year old loblolly pine (Pinus taeda) plantations as affected by moisture and nutrient availability. Can. J. For. Res. 30, 347 359.

PAGE 232

232 McCarthy, D.R., Brown, K.J., 2006. Soil respiration responses to topography, canopy cover, and prescribed burning in an oak hickory forest in southeastern Ohio. For. Ecol. Manage. 237, 94 102. Medvedeff C. 2012. The effect of an extreme restoration approach on microbial carbon cycling in a restored subtropical wetland. Ph.D. Dissertation, University of Florida, Gainesville, Florida. p. 167. Meigs, G., Donato, D., Campbell, J., Martin, J., Law, B., 2009. Forest fire i mpacts on c arbon u ptake, s torage, and e mission: t he r ole of b urn s everity in the Eastern Cascades, Oregon. Ecosystems 12, 1246 1267. Menges, E., Gordon, D., 2010. Should mechanical treatments and herbicides be used as fire surrogates to manag Fla. Sci. 73, 147 174. Michelsen Correa, S., Scull, P., 2005. The impact of reforestation on soil temperature. Middle States Geographer 38, 39 44. Miller, S., Wade, D., 2003. Re introducing fire at the urban/wild land interfa ce: planning for success. J. For. 76, 253 260. that sustains: the nexus between silviculture, frequent prescribed fire, and conservation of biodiversity in longleaf p ine forests of the southeastern United States. Can. J. For. Res. 36, 2724 2736. McKinley, D., Ryan, M., Birdsey, R., 2011. A synthesis of current knowledge on forests and carbon storage in the United States. Ecol. App. 21, 1902 1924. Moser, W.K., 2002. Ex amination of stand structure on quail plantations in the Red Hills region of Georgia and Florida managed by the Stoddard Neel system: an example for forest managers. J. For. 75, 443 449. Myers, R.L., Ewel, J.J., 1990. Eco systems of Florida. University of C entral Florida Press, Orlando, Florida. Nadelhoffer, K.J., Boone, R.D., Bowden, R.D., Canary, J.D., M icks, P., Ricca, A., Aitkenhead, J.A., Lajtha, K., Mcdowell, W.H., 2004. The DIRT Experiment and root influences on forest soil organic matter stocks and function, in: Aber, J., Foster, D. (Eds.), Forests in Time. Yale University Press, New Haven, CT, pp. 300 315. Neary, D., Klopatek, C., DeBano, L., and Ffolliott, P., 1999. Fire effects on belowground sustainability: a review and synthesis. For. Ecol. Manage. 122, 51 71. Nobili, M.D., Contin, M., Mondini, C., 2001. Soil microbial biomass is trigge red into activity by trace amounts of substrate. Soil Biol. Biochem. 33, 1163 1170.

PAGE 233

233 Noormets, A., Gavazzi, M.J., Mcnulty, S.G., Domec, J., Sun, G., King, J.S., Chen, J., 2010. Response of carbon fluxes to drought in a coastal plain loblolly pine forest. Gl ob. Change Biol. 16, 272 287. Outcalt, K.W., Wade, D ., 1999. The value of fuel management in reducing wildfire damage, in: Neuenshwander, L., Ryan, K., Gollberg, G., Greer, J. (Eds.), Crossing the Millennium: Integrating Spatial Technologies and Ecological Principles for a New Age in Fire Management Volume II. University of Idaho and the International Association for Wi ldland Fire, Boise, Idaho, pp. 271 274. Outcalt, K., Wade, D., 2004. Fuels management reduces tree mortality from wildfires in southeaster n United States. South. J. Appl. For. 28, 28 34. Paisley, C., 1989. The Red Hills of Florida, 1528 1865. University of Alabama Press, Tuscaloosa Alabama Post, W., Emanuel, W., Zinke, P., Stangenberger, A., 1982. Soil carbon pools and world life zones. Na ture 298, 156 159. Powell T.L., Gholz, H., Clark, K. Starr, G., Cropper, W., Martin, T. 2008. Carbon exchange of a mature, naturally regenerated pine forest in north Florida. Glob. Change Biol. 2523 2538. Quinn Thomas, R., Canham, C.D., Weathers, K.C., Goodale, C.L., 2009. Increased tree carbon storage in response to nitrogen deposition in the US. Nature Geosci. 3, 13 17. Raich, J.W., Schlesinger, W.H., 1992. The global carbon dioxide flux in soil respiration and its relationship to vegetation and climat e. Tellus B 44, 81 99. Raich, J.W., Tufekciogul, A., 2000. Vegetation and soil respiration: Correlations and controls. Biogeochemistry 48, 71 90. Reinke, J., Adriano, D., McLeod, K., 1981. Effects of litter alteration on carbon dioxide evolution from a Sou th Carolina pine forest floor. Soil Sci. Soc. Am. J. 45, 620 623. Robertson, K.M., Ostertag, T.E., 2007. Fuel characteristics and fire behavior predictions in native and old field pinelands in the Red Hills Region, southwest Georgia, in: Proceedings of the 2nd International Wildland Fire and Management Congress, Orlando, Florida. Robbins, LE, and RL Myers. pp. 109 120. Reid, A.M., Robertson, K.M., Hmielowski, T.L., 2012. Predicting litter and live herb fuel consumption during prescribed fires in native and old field upland pine communities of the southeastern United States. Can. J. For. Res. 42, 1611 1622.

PAGE 234

234 Reynolds, B.C., Hunter, M.D., 2001. Responses of soil respiration, soil nutrients, and litter decomposition to inputs from canopy herbivores. Soil Biol. B iochem. 33, 1641 1652. Rustad, L.E., Huntington, T.G., Boone, D., 2000. Controls on soil respiration : Implications for climate change. Biogeochemistry 48, 1 6. Ryan, M.G., Law, B.E., 2005. Interpreting, m easuring, and m odeling s oil r espiration. Biogeochemistry 73, 3 27. Ryu, S., Concilio, A., Chen, J., North, M., Ma, S., 2009. Prescribed burning and mec hanical thinning effects on belowground conditions and soil respiration in a mixed conifer forest, California. For. Ecol. Manage. 257, 1324 1332. Saiz, G., Byrne, K., Butterbach Bahl, K., Kiese, R., Blujdea, V., Farrell, E.P., 2006. Stand age related effec ts on soil respiration in a first rotation Sitka spruce chronosequence in central Ireland. Glob. Change Biol. 12, 1007 1020. Salamanca, E., Raubuch, M., Joergensen, R., 2006. Microbial reaction of secondary tropical forest soils to the addition of leaf lit ter. Appl. Soil Ecol. 31, 53 61. Samuelson, L., Mathew, R., Stokes, T., Feng, Y., Aubrey, D., Coleman, M., 2009. Soil and microbial respiration in a loblolly pine plantation in response to seven years of irrigation and fertilization. For. Ecol. Manage. 258 2431 2438. Samuelson, L.J., Johnson, K., Stokes, T., Lu, W., 2004. Intensive management modifies soil CO 2 efflux in 6 year old Pinus taeda L. stands. For. Ecol. Manage. 335 345. Sapronov, D., Kuzyakov, Y., 2007. Separation of root and microbial respirati on: Comparison of three methods. Eurasian Soil Sci. 40, 775 784. Samuelson, L.J., Whitaker, W.B., 2012. Relationships between soil CO 2 efflux and forest structure in 50 year old longleaf pine For. Sci. 58. 472 484. Sayer, E.J., 2006. Using experimental ma nipulation to assess the roles of leaf litter in the functioning of forest ecosystems. Biol. Rev. Camb. Philos. Soc. 81, 1 31. Schlesinger, W., Andrews, J., 2000. Soil respiration and the global carbon cycle. Biogeochemistry 48, 7 20. Shan, J., Morris, L. a., Hendrick, R.L., 2002. The effects of management on soil and plant carbon sequestration in slash pine plantations. J. Applied Eco. 38, 932 941. Stoddard, H.S., 1969. Memoirs of a Naturalist, 2 nd E d. University of Oklahoma Press, Norman, Oklahoma.

PAGE 235

235 Subke, J ., Inglima, I., Cotrufo, M.F., 2006. Trends and methodological impacts in soil CO 2 efflux partitioning : a meta analytical review. Glob. Change Biol. 12, 921 943. Subke, J., Voke, N.R., Leronni, V., Garnett, M.H., Ineson, P., 201 1 Dynamics and pathways of autotrophic and heterotrophic soil CO 2 efflux revealed by for est girdling. Ecology 99, 186 193. Sulzm an, E.W., Brant, J.B., Bowden, R.D., Lajtha, K., Biogeochemistry, S., Mar, S.R., Brant, B., 2012. Con tribution of aboveground litter, belowground litter and rhizosphere r espiration to total soil CO 2 efflux in an old growth coniferous forest. Biogeochemist ry 73, 231 256. Tang, J., Qi, Y., Xu, M., Misson, L., Goldstein, A.H., 2005. Forest thinning and soil respiration in a ponderosa pine plantation in the Sierra Nevada. Tree Physiol. 25, 57 66. Varner, J.M., Gordon, D.R., Putz, F.E., Hiers, J.K., 2005. Resto ring f ire to l ong u nburned P inus palustris e cosystems: n ovel f ire e ffects and c onsequences for l ong u nburned e cosystems. Restor. Ecol. 13, 536 544. Vince, S.W., Duryea, M.L., Macie, E.A., Hermansen, L.A., 2005. Forests at the Wildland Urban Interface. CRC Press, Boca Raton, Florida Waldrop, T., Goodrick, S., 2012. Introduction to prescribed fire in Southern ecosystems. Science Update SRS 054. US Department of Agriculture, Forest Service, Southern Research Station, Asheville, NC. 80 p. Waldrop, T.A ., White, D.L., Jones, S.M., 1992. Fire regimes for pine grassland communities in the southeastern United States. For. Ecol. Manage. 47, 195 210. Wang, C., Yang, J., Zhang, Q., 2006. Soil respiration in six temperate forests in China. Glob. Change Biol. 12, 2103 211 4. Warren, J.M., Iversen, C.M., Garten, C.T., Norby, R.J., Childs, J., Brice, D., Evans, R.M., Gu, L., Thornton, P., Weston, D.J., 2012. Timing and magnitude of C partitioning through a young loblolly pine (Pinus taeda L.) stand using 13 C labeling and shad e treatments. Tree Physio. 32, 799 813. Way, A.G., 2006. The Stoddard Neel method: forestry beyond one generation. For. Hist. Today 16 23. Wertin, T.M., Teskey, R.O., 2008. Close coupling of whole plant respiration to net photosynthesis and carbohydrates. Tree Physiol. 28, 1831 40. Whitaker, W.B., 2010. Relationships between forest structure and soil CO 2 efflux in 50 year old longleaf

PAGE 236

236 White, D., Waldrop, T., Jones, S., 1990. Forty years of prescribed burning o n the Santee Fire Plots: effects on understory vegetation, in: Nodvin, S., Waldrop, T. (Eds.), Fire and the Environment: Ecological and Cultural Perspectives: Proceedings of an International Symposium. US Department of Agriculture, Forest Service, Southeas tern Forest Experiment Station, Knoxville, TN, pp. 51 59. Wiedinmyer, C., Neff, J.C., 2007. Estimates of CO 2 from fires in the United States: implications for carbon management. Carbon Bal. Mgt. 2, 10. Wiedinmyer, C., Hurteau, M.D., 2010. Prescribed fire a s a means of reducing forest carbon emissions in the western United States. Environ Sci. Technol. 44, 1926 1932. Williams, R.J., Hallgren, S.W., Wilson, G.W.T., 2012. Frequency of prescribed burning in an upland oak forest determines soil and litter proper ties and alters the soil microbial community. For. Ecol. Manage. 265, 241 247. Woodbury, P.B., Smith, J.E., Heath, L.S., 2007. Carbon sequestration in the US forest sector from 1990 2010. For. Ecol. Manage. 14 27. Xu, J., Chen, J., Brosofske, K., Li, Q., W eintraub, M., Henderson, R., Wilske, B., John, R., Jensen, R., Li, H., Shao, C., 2011. Influence of timber harvesting alternatives on forest soil respiration and its biophysical regulatory factors over a 5 year period in the Missouri Ozarks. Ecosystems, 14 1310 1327. Zak, D.R., Holmes, W.E., Finzi, A.C., Norby, R.J., Schlesinger, W.H., 2003. Soil nitrogen cycling under elevated CO2: A synthesis of forest FACE experiments. Ecol Appl 13, 1508 1514. Zhou, X., Zhou, L., 2012. Temperature sensitivity of respi ratory processes vary across scales. Ecosys Ecograph. 2, 10 11. Zimmermann, M., M eir, P., Bird, M., Malhi, Y., C ahuana, A., 2009. Litter contribution to diurnal and annual soil respiration in a tropical montane cloud forest. Soil Biol. Biochem. 41, 1338 1 340.

PAGE 237

237 BIOGRAPHICAL SKETCH David Robert Godwin grew up in Tallahassee, Florida A life long naturalist with a keen appreciation for maps, he earned his Bachelor of Science in g eography from Florida State University in 2003 After discovering an interest in forest management and fire ecology while working on public wildlife management areas, he started graduate school in 2007 and in 2008 completed a Master of S cience degree from the School of Forest Resources and Conservation at the University of Florida. He began his d octorate in 2009 under the tutelage of Dr. Leda Kobziar at University of Florida and successfully defended his dissertation in 2012. David is married to Katie ( Kight ) Godwin of Orange Pa rk and St. Augustine, Florida. Their delightful son Ben jamin was born in Jacksonville Florida in 2012. They currently reside in St. Augustine, Florida.